Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Fork point: upstream post-8.1.1 (2025-11-27) Upstream baseline: WUD 8.1.1 + 65 merged PRs on main (Vue 3 migration, Alpine base image, Rocket.Chat trigger, threshold system, semver improvements, request→axios migration, and more)

Unreleased

1.7.0-rc.18 — 2026-10-04

Security

Login lockout now keys on the credentials actually presented. A request could previously name a different username in its body to dodge an account's failed-login lock, leaving only the per-IP limit.
Container start, stop and restart responses now redact sensitive environment values the same way the container detail endpoint does.
The demo site now sends a full Content-Security-Policy. apps/demo/vercel.json sent only frame-ancestors, so script-src and default-src fell open and ZAP raised rules 10055-4, 10055-5 and 10055-13 on every scan. The policy now pins scripts, the mock service worker and form targets to 'self', and allows only the jsDelivr and Iconify hosts the mock icon and font handlers fetch from.
Add Cross-Origin-Opener-Policy: same-origin and a static strict CSP on /api responses to the website, and stop sending X-Powered-By.

Changed

ZAP alerts in code scanning are now keyed per scanned site. scripts/zap-json-to-sarif.mjs stripped the origin from every location, so getdrydock.com, the demo and the app scan shared one alert per rule and path, and dismissing it for the public site hid the same finding on the app. Locations now read getdrydock.com/robots.txt. Expect one round of reopened and closed alerts when the next scans upload.
The weekly DAST scans of getdrydock.com and the demo can now pass. The two public-site ZAP jobs read a new .zap/rules-public-site.tsv, which adds IGNORE entries for the reviewed false positives (SQL, private IP, timestamp, eval, debug-error and application-error text matched in docs prose, plus proxy, user-agent and public-file CORS notices) and the accepted CORP and COEP choices. The app scan in ci-verify.yml keeps the stricter .zap/rules.tsv, and a workflow test pins both.
A stable release shows its own version instead of the release candidate it was promoted from. A stable release is the last release candidate's image, promoted unchanged, so the 1.6.1 image reported 1.6.1-rc.15 in the UI, the API and Home Assistant. Drydock now shows the base version (1.6.1) and keeps the full build identity as a separate build field. The build appears in the About dialog, under Config > General and in the agent detail panel when it differs from the version, and the startup banner (shown on a TTY) reads version 1.6.1 (build 1.6.1-rc.15). The drydock is starting log line carries no version. GET /api/v1/app and GET /api/v1/agents keep version as the base version and add build, the agent dd:ack event carries both, the debug dump adds drydockBuild next to drydockVersion, and Home Assistant's sw_version and the OpenAPI document version use the base version. API clients that parsed a prerelease suffix out of version should read build instead. Values that aren't a semver with a prerelease suffix, such as local or ci, are reported unchanged. The controller applies the same split to the version an older agent or a Portwing edge agent reports, so those agents show their base version too. Reported in #1284.

Fixed

The website docs pages /docs/v1.6/changelog, /docs/v1.6/configuration/ui, /docs/v1.7/changelog and /docs/v1.7/configuration/ui no longer return a 500. A remark plugin renders bare {column}, {date} and {countdown} in prose as literal text instead of evaluating them as JS expressions.
Stop the dashboard from showing a connected Local Docker host when no local watcher is configured. Agent-only fleets show only their configured agents.
Apply the configured outbound HTTP timeout to Docker Hub publish-date metadata requests, so a stalled response cannot indefinitely hold up container discovery. Existing bounded retries and publish-date failure handling are unchanged.
Keep bulk scan counts and scanner results accurate when a post-scan notification fails. Each completed task is counted once, and notification delivery failures no longer replace a completed scan's status with an error.
Keep bulk scan progress accurate for large fleets and busy event streams while HTTP acceptance is delayed. Fresh request correlation and server-owned cumulative counts replace the 500-entry early-event buffer, with bounded client state and explicit recovery for conflicting or lost progress. Duplicate/replayed events do not overcount, and failed tasks still advance progress.
Correlate bulk scan progress with the accepted scan cycle, retain early completion events, and ignore duplicate or unrelated scans. The Security page now shows localized request/progress errors and supports an explicit retry without automatically repeating a scan request. Lost progress requires a successful read-only results refresh before starting another scan; refreshing does not establish whether the original scan has finished.
Release-gated store migrations now compare against the base version, so a future migration gated on a release runs on that release's stable image rather than one release later.

1.7.0-rc.17 — 2026-10-02

Security

Patch brace-expansion to 5.0.12 and the website to Next.js 16.3.6. Update DOMPurify to 3.4.16 for its in-place sanitization fix.
Update Undici to 8.10.2, Nodemailer to 10.0.9, gRPC to 1.14.5, Moment to 2.31.0, fast-uri to 4.1.5, and ip-address to 10.7.1 for the newly published dependency advisories. Keep UI and E2E Undici on their patched 7.29.1 line. Nodemailer 10 requires Node.js 20 or newer; Drydock already requires Node.js 24.
Update Alpine OpenSSL to 3.5.9-r0 after 3.5.8-r0 left the package index and the pinned install stopped resolving.
Patch Alpine zlib 1.3.2 with the upstream fix for CVE-2026-85091 and update libexpat to 2.8.5-r0 for CVE-2026-93990. The temporary zlib APK retains its upstream version and records a unique local revision; its exact backport is documented in the image scanner's VEX evidence.

Fixed

#1284: a disabled healthcheck no longer health-gates the update. A container with healthcheck: disable: true (stored by Docker as Test: ["NONE"]), an empty test, or a healthcheck block with only timing fields was treated as having a healthcheck. Docker never reports a health state for those, so the Docker action waited out the whole rollback window and rolled back an update that had started fine. The gate now applies only when the container has a real CMD or CMD-SHELL probe or Docker is reporting health.
#1280: update policy set in the UI survives recreation of an agent-managed container. An agent reports a recreated container as a removal of the old id and an addition of the new one. The controller deleted the old row without keeping its policy, so a maturity setting made in the UI was lost every time the container was updated. The policy is now carried to the replacement whichever of the two events arrives first, and also when a reconnecting agent's snapshot drops the old id before listing the new one. It is only ever carried to a container with the same agent, watcher and name. A local scan that catches Drydock's temporary -old-<timestamp> rename no longer stores that name, which could make the same carry-over miss on the controller host.
#1281: the Docker Hub config blob redirect is no longer logged as a failure. Registry redirects are not followed, so the optional created-date lookup against Docker Hub always ends in a 307. The debug line now says the created date is optional and was skipped, instead of "Unable to fetch image config blob created date", which read like the reason an update had not run.
Let accepted bulk vulnerability scans finish after the HTTP request completes normally. Previously, inventories larger than the four-scan concurrency limit could stop after the first batch while the UI kept waiting for the remaining results. Prematurely closed, incomplete requests still stop queued scans.

1.7.0-rc.16 — 2026-09-15

Fixed

Keep Crowdin source uploads and translation PRs on the highest integration branch, including runs triggered by maintenance-line pushes, so the shared translation branch cannot target an older release line with newer product changes.
Accept newer stable YAML and PostHog pins in dependency guards while enforcing the YAML security floor and manifest/lockfile consistency, including nested YAML installs.
Update the Docker image's timezone package pin to tzdata=2026d-r0, available in Alpine 3.24 for amd64 and arm64, after 2026c-r0 left the package index.
Backport #1139 to publish agent container state on the MQTT topic Home Assistant discovery advertises, fixing a cause of entities staying Unknown with v1.7's default agent segmentation. Existing retained state on the old unscoped topics is not removed automatically.
Restore website referral-source reporting while keeping referrer URLs and campaign parameters out of analytics.

1.7.0-rc.15 — 2026-09-10

Fixed

Stop an edge agent's in-flight component initialization from publishing after disconnect and replacing its reconnected owner. Retired registrations clean up only their own components. Disconnected Docker proxy, log, delete, and exec requests now fail immediately without creating new request state or sending frames.

1.7.0-rc.14 — 2026-09-08

Fixed

A maintenance cut labeled the shipped image with main's commit instead of the commit it was built from. Both docker/metadata-action steps in release-cut.yml left org.opencontainers.image.revision at its default, github.sha, which is the workflow run's own checkout rather than the dev-branch source commit the build used. The v1.6.1-rc.9 staging image carried 5ae315227 in that label instead of 2969675ef. Both steps now set the label from the release source SHA.
Rolling back a container (auto-rollback on an unhealthy update, or a manual restore) now runs the same runtime-config sanitization as an update, so an entrypoint or command the newer image introduced is no longer copied onto the older image and the rolled-back container starts.
Tag family matching now requires the candidate's variant suffix to match the current tag's exactly, so a container on 1.27.3-alpine is no longer offered 1.28.0-alpine-perl (or 1.28.0-alpine from -alpine-slim) under dd.tag.family=loose or a dd.tag.include filter.
Precision-only suffix carve-out now requires at least one digit placeholder, so a trailing dot (1.3.0-alpine.) no longer counts as the same tag family as 1.3.0-alpine.
Monthly and longer watcher schedules could expire scans after 1 ms. The scan deadline was twice the cron interval, which overflowed Node's timer limit and cleared the in-flight scan guard almost immediately. Deadlines now stop at the largest supported delay, preserving the existing ten-minute floor and shorter schedule behavior.

1.7.0-rc.13 — 2026-09-08

Fixed

Containers on a floating tag that drydock first saw before v1.5.0-rc.17 could stay marked Current forever, even when the registry had a newer digest. image.digest.watch was written once at first discovery and never revisited, so a row discovered before the digest-watch default changed from !isDockerHubDomain(domain) to "watch when the tag is meaningful" (v1.5.0-rc.17) kept the old false default permanently; only recreating the container picked up the new one. It is now re-derived every scan, the same way isLocalImage and digest.repoDigests already are, and an explicit dd.watch.digest label or imgset override still wins. (#1070)
The weekly ZAP full scan of getdrydock.com ran into its 60-minute job timeout on every run, so it never produced a report. The scan step now caps the spider at 10 minutes and the active scan at 35 minutes (5 per rule), leaving room for startup, the passive scan and the report inside the job budget; a workflow test pins the arithmetic.

1.7.0-rc.12 — 2026-09-06

Fixed

The demo site did not send Cross-Origin-Opener-Policy, so the weekly DAST scan failed on ZAP rule 90004 every run. apps/demo/vercel.json now sends same-origin next to the existing Cross-Origin-Embedder-Policy header.
The arm64 pass of the image arch check failed on every multi-platform cut with docker: cannot overwrite digest sha256:<index>. scripts/check-image-arch.sh ran once per platform against the same index-digest reference (ghcr.io/codeswhat/drydock:release-staging-N@sha256:<index>), and docker's classic image store cannot hold two platform variants under one digest, so the amd64 pass succeeded and the arm64 pass that followed it always failed. This killed the v1.6.1-rc.9 cut. The script now resolves each platform's own manifest digest out of the index via docker buildx imagetools inspect --raw and jq before it runs docker, skipping attestation entries, and falls back to running the reference unchanged when it isn't an index at all.
Two more prunes cleared a container's update policy the same way the startup prune used to. The startup prune stashes a departing record's update policy under its Docker id so the record that replaces it inherits it, but the agent-removal prune and the agent's own stale-container prune still called deleteContainer plainly. An agent removed from config or renamed, or a container handed from one agent to another (or back to the controller's own watcher), lost its snooze, maturity mode and minimum age, and skipped tags or digests the same way the startup case did before that fix. Both prunes now pass identityChangeExpected: true as well, so the same Docker-id stash carries the policy across either hand-off.
A manual "check now" (the dashboard, the API, a webhook, or the controller polling an agent) could fire the same notification twice if it landed while a scan was already running. watch() has no re-entrancy guard of its own, so POST /api/v1/containers/watch, POST /api/v1/webhook/watch, and the agent's own watcher API each started a second, fully independent scan on top of one already in progress from the cron schedule or another manual request, and under once=true both passes could read "not yet notified" before either had finished writing its own history, so a trigger like Slack or Telegram sent the same update twice. All three now route through the same single-flight scan orchestration the cron schedule uses: a call that lands mid-scan is folded into that scan's one follow-up instead of starting an independent scan, and the response reports it (result.coalesced in the JSON body for the first two, an X-Drydock-Watch-Coalesced header for the agent endpoint, which keeps its existing bare-array body for compatibility across controller/agent version skew). Manual and API scans still run regardless of a configured maintenance window, exactly as before; only automatic installation is deferred by it. Separately, handleMaturityGateClearedEvent's own once=true check read notification history with a bare, unreserved lookup, so two overlapping evaluations of the same maturity-cleared event could both pass it before either recorded a result; it now takes the same reservation the generic update-available path does before dispatching.
A once-notification reservation that never settled held its dedup key for the process lifetime. once=true reserves a triggerId::containerId::eventKind::resultHash key before sending so an overlapping evaluation of the same result cannot send twice, and releases it once the send settles. runHandlerWithTimeout detaches a handler that misses its 30-second deadline instead of waiting on it forever, so a provider that never resolves or rejects left its reservation held forever too, and every later scan for that exact result read it as still in flight and silently skipped the send. The reservation now carries its own expiry, four times the handler timeout, that releases it with a warn log naming the key if nothing released it first.
DR-121: the session store and the main store wrote the same /store/dd.json, and whichever one saved last erased the other's data. express-session's connect-loki store opened its own independent LokiJS instance on the exact file the main store already used, and LokiJS's saveDatabase() always serializes the whole in-memory database, so a session autosave (every 5 seconds, and armed by nothing more than an authenticated request touching its session) reverted every container, setting, and audit row the main store had written since boot, while the main store's own autosave (every 5 minutes) deleted the Sessions collection out from under active logins. In practice this meant containers or settings written after startup could vanish from dd.json after a crash or a hard stop, and a restart could log every user out. The session store now writes to its own sibling file, dd-sessions.json by default (derived from DD_STORE_FILE, so a custom filename still gets a distinct sibling rather than colliding with anything else on the volume), and the main store drops and logs a stale Sessions collection left behind in dd.json by an older build instead of continuing to re-save it. Backups remain scoped to the main store file; sessions are not included and are expected to be dropped on restore, which just logs everyone out rather than corrupting anything.

Documentation

The agents page's paired Gitea registry example had the controller talking HTTPS to an agent serving plain HTTP. The controller block set DD_AGENT_REMOTE1_CAFILE=/certs/agent-ca.pem, but the agent block above it had no DD_SERVER_TLS_* variables or certificate mounts, so the example copied as written could never connect. The agent block now mounts agent.pem/agent-key.pem and sets DD_SERVER_TLS_ENABLED, DD_SERVER_TLS_CERT, and DD_SERVER_TLS_KEY, with a comment noting the certificate must be signed by the agent-ca.pem the controller mounts and be valid for the host the controller dials.

1.7.0-rc.11 — 2026-09-05

Fixed

OIDC login on v1.7.0-rc.1 through rc.10 bounced straight back to the login page. The service worker's navigation fallback denied /api/ and nothing else, so every other top-level navigation was answered from the precached app shell. The identity provider redirects the browser to /auth/oidc/<name>/cb?code=..., which is a document navigation, so the callback was served index.html and never reached Express: no code exchange, no session, and the SPA booted and bounced to /login. Basic auth was unaffected because it authenticates over fetch, which the navigation fallback never touches. skipWaiting and clientsClaim re-register the worker on the next load, so clearing site data only helped until the page reloaded. The navigation fallback now skips every server-owned route (/api, /auth/, /health, /metrics) and serves the shell only for the SPA's own paths; the /auth settings view still loads from the shell, because Express matches that mount path too but has no handler for it, so the shell answers instead. v1.6 was never affected, it shipped no service worker. (#939)
A controller running its own default local watcher refused every container reported by an agent whose watcher was also named local. The ownership gate added in v1.7.0-rc.6 (#922) decided ownership of a container that had no store row yet from the watcher's name: the report was refused whenever the controller had a watcher registered under that same name. The race it closes is real, an agent can otherwise claim a container id that lives on the controller's own host before the controller's watch cycle writes the record, but a watcher name is not evidence of who owns an id, and it collides by default. A controller with no DD_WATCHER_* registers its default watcher as local, and the agents quickstart sets DD_WATCHER_LOCAL_SOCKET on the agent, so both sides are called local while watching entirely different hosts. On a fresh store that refused every container the agent reported, on every ingest path, logging watcher 'local' belongs to the controller's local watcher namespace and leaving zero agent rows in the UI and the API, permanently. An install that already had the rows stored kept working until a container was recreated, because the gate also feeds the prune keep-set: the stale row was pruned as a removal rather than a replacement and the new id was refused, losing that container for good. Ownership of a no-record id is now decided by what the controller's own watchers have actually enumerated. Each controller-local Docker watcher records the container ids its listContainers() call returns, replacing the set every cycle so a container that leaves the controller's host stops being claimed, and clearing it when the watcher is deregistered. An agent report with no store row is refused only when its id is in one of those sets, and the log line names the watcher holding it. A controller-side agent watcher running the Docker transport records nothing, since it enumerates the agent's daemon rather than the controller's. Disabling the controller's local watcher or renaming the watcher on either side were the only workarounds.
A rollback of a compose-managed container redeployed the update it was undoing. Dockercompose.recreateContainer wrote the image it was handed into the compose file and then handed the runtime refresh no image at all, so the refresh re-derived one from the container's own update candidate. The compose file said the backup and the container that came up ran the update, while the API answered Container rolled back successfully and the audit row recorded a success. Clearing the update result did not help either, because the fast resync that follows an update had already moved the container's tag to the new version by the time anyone rolled back. The recreate now passes the caller's image through to the refresh, so the container, the compose file and the report agree. Automatic rollback on a failed healthcheck never got that far on a compose stack at all: the health monitor called the trigger with only the container's id and name, and the compose action read the registry, the watcher and the compose labels off the container to find the service to rewrite, so it threw a TypeError before it reached the compose file. That failure was logged as an auto-rollback failure and recorded in an auto-rollback audit row carrying the error, neither of which stopped the unhealthy container from staying up on the version that had just failed its healthcheck. The monitor now hands over the whole container, carrying the id of the replacement it is rolling back, and the container type it arrives as requires the registry and the watcher, so the stand-in that caused this cannot compile again. Getting that far exposed the next one: the monitor logged that it was pulling the backup image and never pulled it. Neither recreate fetches the image for itself, because the manual rollback in app/api/backup.ts pulls before it calls in and a second pull there would be wasted work, so a backup image no longer on the host failed at create, on the Docker path after the running container had already been stopped and removed and on the compose path after the runtime refresh removed it and then "restored" it onto the failing update the rollback existed to undo. The automatic rollback now pulls the backup reference, digest-pinned when the record carries one, before it touches anything, and the compose refresh refuses an image it cannot find locally before the stop and remove rather than at create, so a rollback that cannot happen leaves the running container where it is. That pull is skipped when the backup image is already on the host, so a rollback on a host with no route to the registry, or one hitting a Docker Hub anonymous rate limit, still restores from the image the prune retained.
The arm64 image published for v1.7.0-rc.4 through rc.10 was an x86-64 image wearing an arm64 label. #881 rolled the node:24-alpine and alpine:3.24 base pins to digests that name a single amd64 manifest instead of the multi-arch image index, and buildx resolves a digest pin the same way for every --platform, so the arm64 stage built on an amd64 rootfs and nothing in the build, the manifest list, or the image scans noticed. On a Raspberry Pi the container died at exec /sbin/tini: exec format error (#1021). Both pins are back on image index digests, CI now rejects any FROM pinned to a per-platform manifest, and the release reads the ELF machine type of /sbin/tini, /usr/local/bin/node and /bin/healthcheck in each platform of the image it built and refuses to sign, tag, or promote one whose binaries do not match the platform they are published under.
A container that moved to an agent stayed stranded when the controller's local watcher was turned off. DD_LOCAL_WATCHER=false leaves the controller with no local watcher registered, and the startup prune that clears records naming a watcher that no longer exists skipped that case entirely, so the old controller-owned record survived every restart. The agent's report for the same container was then refused as owned by the controller, the record was never rewritten to name the agent, and the container sat on whatever it last said. The prune now runs when no local watcher was configured in the first place. It still leaves the store alone when watchers were configured and all of them failed to register, which looks identical in the registry but means something transient rather than an operator decision. A snooze, a maturity mode and any skipped tags or digests set from the UI live on that record, and they travel with the container rather than dying with it: the prune stashes them under the container's Docker id and the agent's copy inherits them when it arrives.
A container handed from the controller to an agent came up with its update policy cleared. The startup prune deletes the controller's record so the agent can claim the container, and it deleted it the way it deletes a container that is gone for good. Nothing would have helped if it had asked for a replacement either: the stash that carries an update policy across a delete is keyed on agent plus watcher plus name, and a move to an agent changes all three, so the incoming record looked up a key nothing had ever written. A snooze, a maturity mode and minimum age, and skipped tags or digests were all silently back to defaults, which for a maturity gate means the next update goes out with no soak at all. The prune now stashes the policy under the container's Docker id, which is the one thing the move leaves alone because it is the same physical container on both sides, and a record arriving under a new identity with that id inherits it. The identity-keyed stash is untouched and still covers the opposite move, a recreate, where the id changes and the identity does not. The same fix covers a watcher rename, which loses the policy for the same reason. A container recreated as part of the move mints a new Docker id and matches neither key, so that one still starts from the declarative policy.
One local watcher failing to register deleted its containers' records when a second watcher registered fine. The startup prune keeps records whose watcher is in the registry, and #1025 only protected the case where every configured watcher failed. With two configured and one of them broken by an unreadable CA file, a socket that is not there, or a value the schema rejects, the surviving watcher keeps the registered set non-empty, so the broken watcher's records were pruned as though the operator had renamed it away, and it came back to an empty slate once its configuration was fixed. The prune now keeps any record whose watcher is still present in DD_WATCHER_*, whether or not it registered this run, and only prunes a watcher that is neither registered nor configured. registerWatchers also waits for every registration to settle instead of returning the moment one rejects, so the prune reads the registry the run actually produced rather than whichever watchers happened to have landed by the time another one failed.

Documentation

The agents page didn't say registries have to be configured on every agent, not just the controller. A traditional agent runs its own watcher and does its own registry matching and update checks, so DD_REGISTRY_* configured only on the controller left every agent-reported container from that registry stamped unknown, credentials are never pushed from controller to agent. The controller needs the same registry configured too, or the container's registry link in the UI resolves to a registry the controller was never told about. A new "Registries on agents" section spells this out with a worked Gitea example on both sides. Reported in #945.
The agents page's registries example pointed DD_AGENT_REMOTE1_CAFILE at a CA file the controller's compose service never mounted. The environment variable named /certs/agent-ca.pem, but the example's services.drydock had no matching volumes: entry, so a reader following it as written couldn't find the file on the container's filesystem. The example now mounts ./certs/agent-ca.pem read-only at that path, and the inline comment points at the controller environment variables reference instead of citing an unrelated plain-HTTP quickstart.
The watchers page said manual updates bypass the maintenance window but didn't say the window gates the entire scheduled scan, not just installing an update. A closed window means new containers stay invisible in the UI until the next window opens, container state shown in the UI goes stale (a container stopped during the last window still shows as stopped, and a start action on it fails because Docker refuses to start a container that's already running), and update notifications are deferred right along with the update itself. A manual scan (POST /api/v1/containers/watch, or the UI) bypasses the window the same way a manual update does, because it calls the watcher's watch() directly rather than the cron path that checks it. Discussed in #946.

1.7.0-rc.10 — 2026-09-04

Fixed

An update could be announced twice, and a torn-down watcher could still warn about a scan deadline it no longer owned. The once=true reservation added for #972 covered the simple notification path only, so batch and digest eligibility still did a plain read of the notification history, and that history is written after the send resolves. A manual single-container scan overlapping a cron scan therefore passed the check in both evaluations and sent the same candidate twice in batch mode, and in digest mode a report landing while a flush was still sending re-buffered the same result behind the send, so the post-send delete skipped it and the next flush sent it again. Batch and digest eligibility now take the same synchronous reservation the simple path does, the batch releases every reservation it took in a finally, and the digest flush holds one for exactly the results it is sending. Two paths that bypassed the gate entirely are closed with it. The digest flush swaps the current store container in for the buffered one at send time, and when that substitute was a result an earlier flush had already sent it went out again while the post-send delete quietly dropped the newer candidate the buffer was holding; it is now skipped and evicted instead, and a genuinely pending update re-enters the buffer on the next scan. Entries in the batch retry buffer, which every batch re-sends until one succeeds, took no reservation at all, so two overlapping retries both carried the same entry to the trigger. Separately, the cron scan's deadline timer lived only in the closure that raced it, so deregistering a watcher while its watch() was stalled left the timer running to the full deadline and then logged "exceeded its deadline" against a watcher that no longer existed, while the caller that started the scan and every caller coalesced into it stayed pending. The timer and a cancel handle now live on the watcher state, deregistration clears the timer and settles the race with its own sentinel so every caller resolves to an empty result with no deadline warning, and the per-scan cleanup is identity-guarded so a late settlement cannot clear a newer scan's handle. A scan requested after teardown is now refused outright, which also covers the docker-events debounce: just-debounce exposes no cancel, so its pending timeout still fires up to five seconds after the watcher is gone and used to start a full scan against it.

1.7.0-rc.9 — 2026-09-03

Security

fflate override added in e2e/ for CVE-2026-45820. @smithy/middleware-compression pins fflate at 0.8.1 exactly, so the fix had to come through an npm override; fflate now resolves to 0.8.3 in e2e/package-lock.json, which is the only workspace that carried the vulnerable range. Lockfile-only change, no runtime code touched.

Fixed

watchFromCron() had no re-entrancy guard, so overlapping scans on a large fleet fired the same trigger multiple times for one update. A full scan can take minutes, and the cron schedule, the docker-events debounce, the discovery-settle timer, and the startup timer could all start one while the previous scan was still running. Each overlapping call ran its own watch(), and a tag first seen mid-burst passed the once=true history check in every one of them before any of them recorded it, so a trigger like Telegram fired once per overlapping scan for the same update. watchFromCron() is now single-flight: a request while a scan is running does not start a second one, it records that a rescan was requested, and exactly one follow-up scan runs once the current scan finishes, so a docker event that arrives mid-scan is not lost. A scan that never settles cannot wedge later ticks either: the in-flight scan is raced against a deadline, and when it fires every caller waiting on that scan, the one that started it and any that coalesced into it, resolves to an empty result and the next tick starts a fresh scan. The "Cron started" log line now also names what triggered the scan (schedule, docker-event, discovery-settle, startup, or maintenance-window). Reported by @tarzan77cz in #972.
A once=true trigger re-fired hours later for a tag update it had already announced, whenever a registry rate-limited the accompanying digest lookup. The notification-history hash for a tag-kind update included the digest and, when the digest was absent, the image's created timestamp. A transient Digest watch failed (429) dropped the digest for that scan only, so created swapped into the hash on the failed scan and back out once the digest lookup next succeeded, and the once history stopped matching an update it had already sent. The hash for a tag-kind update on a container with digest watching configured is now keyed on the tag alone, so a digest lookup failing or succeeding does not change it; a container without digest watching configured keeps the original created fallback, and digest-kind updates stay keyed on the digest. Concurrent evaluations of the same history key are also now atomic within the process: the check and the reservation happen in the same synchronous step, so two overlapping scans evaluating the same result cannot both pass the check and both send, which is why one trigger could fire four times in the same millisecond while another skipped. After upgrading, a container with digest watching enabled, a tag update already pending, and once=true whose stored hash came from a scan where the digest lookup had succeeded re-notifies exactly once, because the stored hash format changed; every scan after that stays stable. Reported by @tarzan77cz in #972.
Two deprecation banner strings in the UI still described the legacy DD_TRIGGER_* env vars and the curl-based healthcheck override as active with a future removal. Both were removed outright in v1.7.0; legacyConfigBody and curlHealthcheckBody (all 17 locales) now say so instead of pointing at a deadline that already passed. The curl banner's {bin} slot also rendered the compose migration snippet with a single $, which compose expands from the host environment before the container starts; doubled it to $${DD_SERVER_PORT:-3000} to match the corrected DEPRECATIONS.md snippet.
The marketing site's Get Started snippets deployed an instance that never became healthy. Neither the docker run quick-start nor the hardened compose preset configured any authentication, so the instance ended up with zero registered auth strategies, which is a deliberate fail-closed state: /health stays 503 forever and there's no way to log in. Both presets also skipped a /store volume, so container state and audit history didn't survive a restart. The quick preset now sets DD_ANONYMOUS_AUTH_CONFIRM=true, the hardened preset sets DD_AUTH_BASIC_ADMIN_USER/HASH, and both mount drydock-store:/store.

Documentation

A docs audit turned up a batch of claims in the README, DEPRECATIONS.md, and the configuration/triggers/registries/API/monitoring/agents docs that no longer matched this tree's code. Compose examples that mangled a pasted argon2id hash and omitted the /store volume; translated READMEs with mistranslated provider names and a backwards roadmap pointer; a deprecation entry using the wrong removal-status label plus four missing DEPRECATIONS.md entries; hook, threshold, and rollback documentation describing behavior the code doesn't have (a major threshold letting digest-only updates through unfiltered, an allowlist that's exact-match rather than basename, a third rollback path that doesn't exist); trigger docs missing template variables and shared config keys; registry docs describing a routing mechanism and a rate-limit scope that don't exist and missing the public.ecr.aws rate row; API docs with a stale audit action list, wrong log-filter semantics, a nonexistent SSE event, and a rate limit bypass that isn't real; and feature docs conflating the docker trigger's health gate with dd.rollback.auto, describing a malformed signing key as a graceful skip when it crashes the controller, and framing a fail-closed /health 503 as a transient startup state. All of it corrected against this tree's actual code, not against v1.8's.

1.7.0-rc.8 — 2026-09-03

Fixed

dd.registry.lookup.image (and its legacy alias dd.registry.lookup.url) did nothing for any container reported by a controller-Docker-transport agent (e.g. Portwing). AgentClient.ts's buildContainerReport() calls normalizeContainer() to pick a registry provider for these agents, but that function reads image.registry.lookupImage as an already-populated field. The controller's own Docker watcher does translate the label, but only when it discovers the container first, and every later agent report replaced the stored image block wholesale, so the field never survived. A container mirrored through a private registry and labeled to divert update checks to its real upstream image (the #336 scenario) silently checked the mirror's own sparse tag copy instead, with no error. buildContainerReport() now applies the same label (falling back to the legacy alias, matching container-init.ts's own precedence) before normalizing, and never overwrites a lookupImage/lookupUrl the agent already reported. Contributed by @hakan42 in #948.
Compose updates ran the post-pull security gate after the service had already been replaced. The Docker-native path pulled, scanned, and enforced policy before stopping the old container, but the Compose path called the same gate only after it had stopped, removed, and recreated the service. A blocked image was therefore already running and already written into the compose file by the time the block verdict arrived. Compose now separates preflight from runtime mutation: it pulls once, captures the exact repository digest, runs signature verification, vulnerability scanning, and SBOM generation against that immutable reference, and only then permits any stop, remove, or create; the per-service platform check still runs before that service is touched. If preflight fails after the candidate compose-file mutation, the existing rollback restores the file before the error is surfaced. Compose-file-once mode preflights every affected service before the first runtime mutation, and a failed preflight terminalizes every queued operation. If the daemon cannot bind a scan to the pulled image, block mode and signature verification fail closed; availability warn records a skipped scan and continues without pretending a mutable-tag scan protected the replacement. Scaled services reuse the same captured identity, private registries are matched exactly, and recovery preserves the operator-facing tag from the hybrid tag@digest reference.
The ordinary Docker update pulled by mutable tag, then verified, scanned and deployed whatever that tag pointed at next. ContainerUpdateExecutor pulled repo:tag and handed the same mutable reference to signature verification, the vulnerability scan and SBOM, the runtime-config compatibility lookup, and finally POST /containers/create. Nothing tied those steps to the image the pull had actually fetched, so a registry retag landing anywhere in that window meant one image was gated and a different one was deployed. The Compose path's equivalent was closed in #952; this is the same hole on the Docker-native path, which is the default for every non-Compose container. The executor now inspects the pulled image once, matches its RepoDigests against the reference's own repository so a lookalike entry for another registry cannot satisfy it, and pins an immutable repo:tag@sha256:... that signature verification, the scan, SBOM, the config lookup and the replacement create all use. Signature verification moved from before the pull into the post-pull gate, because verifying a mutable tag ahead of the pull says nothing about what arrived. If the daemon cannot bind the pulled image to a manifest digest, block mode and signature verification fail closed and terminalize the operation before anything is renamed or created, while availability warn records a skipped scan and continues rather than pretending a mutable-tag scan protected the replacement. Moving the gate behind the pull moved the pre-update hook and the prune/backup step behind the gate with it, so an image the gate goes on to reject no longer fires an operator hook, deletes cached images or writes a rollback row on its way to being refused, and a container failing the gate on every cycle can no longer push a real rollback point out of history. A reference that is already digest-pinned is left alone, the tag the operator configured is still what the operation record, the compose-file sync and the UI show, and the identity binding is now one implementation shared with the Compose path instead of two.
A Docker Hub image pulled through the index.docker.io alias could not be bound to its pulled digest, so the update failed instead of proceeding. The post-pull identity binding matches the daemon's RepoDigests against the repository forms the pulled reference could legitimately be recorded under, and that candidate list knew docker.io and registry-1.docker.io but not the legacy index.docker.io host. The daemon rewrites that alias to docker.io and records Hub images under their short name, so index.docker.io/nginx:1.27 was matched against a candidate list holding only index.docker.io/nginx and never matched the nginx@sha256:... entry it had just pulled. Signature verification and block mode then failed the update closed on an image that was perfectly bindable, and availability warn skipped a scan it could have run. This reaches any container whose registry provider resolves through that host, including a custom registry configured at index.docker.io and a dd.registry.lookup.image diversion that leaves the deploy reference on the alias. Every other Docker Hub alias check in the codebase already covers all three hosts; this one now matches them.
A compose update ran the operator's pre-update hook, pruned images and wrote a rollback row before the security gate had looked at the candidate. Moving signature verification behind the pull put the gate behind those three steps as well, and the compose path kept that ordering when the Docker-native path was changed to defer them alongside the digest pinning. A compose image that failed verification had therefore already fired dd.hook.pre, deleted cached images under PRUNE=true and inserted a backup record on its way to being refused, and a service failing the gate every cycle pushed real rollback points out of history through the backup retention prune that runs after a failed update. The compose trigger now defers the same two steps behind its post-pull gate, except in compose-file-once mode, where the preflight has already gated every service in the batch before the first runtime mutation and there is no post-pull hook left to hang them off. The unbound-image path was the other half of it: when the daemon cannot bind the pulled image to a digest and availability warn allows the update anyway, compose skipped the post-pull hook outright, which under the new ordering would have skipped the pre-update hook and the prune/backup step with it, so it now runs that hook with only the gate half suppressed, exactly as the Docker path already did.
The watcher-snapshot handler still pruned on an empty container list, and a test had frozen that in place. AgentClient has four entry points that ingest a container list, and three of them (handshake(), watch() and handleContainerSync()) already refuse to prune when that list is empty, the first for the cold-start race in #386 and watch() most recently in the rc.6 sweep (#922). handleWatcherSnapshotEvent(), which handles the dd:watcher-snapshot server-sent event, never got one. A reconnecting agent legitimately reports zero containers on this path: filterPendingDiscoveries() only bypasses the discovery-settling delay for container ids already in the agent's own local store, and an agent restart just reset that store, so every genuinely running container is held back on the first cycle. An empty incoming list then made every stored container look like a real removal, so deleteContainer() ran without replacementExpected, stashUpdatePolicyForReplacement() never fired, and the container's maturity policy override was deleted with nothing left to restore it from. This is why the durability fix in #743 did not cover it: that made the stash survive a process restart, and on this path nothing was ever stashed. Note the existing test asserted the broken behaviour outright, having been written from the implementation back in 1.5.0 before the ambiguity was understood, so it is inverted here rather than extended. (#565)
The Trivy supply-chain advisory told readers to expect a version the image stopped shipping. The page names a bundled Trivy version in three present-tense places and quotes the Dockerfile's digest-pinned trivy-bin stage, and all four had drifted: it said 0.72.0 and quoted sha256:cffe3f51… while the image ships 0.73.0 from sha256:7cced7ca…. Caught by running the published rc.6 image and following the page's own instruction, docker exec drydock trivy --version, which is the exact check the advisory tells a reader to run. Someone doing the responsible thing got a mismatch, and the Dockerfile they were invited to compare against on GitHub showed a different digest again, so the only two conclusions available were "the advisory is wrong" or "my image is tampered with". The dated correction paragraphs keep their historical framing rather than being restamped with today's number. The underlying cause was that a Renovate digest bump moves the pin and nothing checked the page against it, so a coupling test now reads the Dockerfile's pin and fails when the advisory disagrees, in the same style as the existing docs version coupling test.
A bad URL on the website left visitors with no way back. There was no not-found.tsx anywhere in the site, so an unknown docs slug fell through to Next's built-in fallback: bare text, no header, no footer, no link into the site. Every real page renders through the marketing or docs shell. Stale docs links are not hypothetical here, the trigger-prefix restructure in v1.5.0 invalidated a batch of them, so the people most likely to hit this were the ones following an old bookmark or a search result. The 404 now renders in the marketing shell with links to the homepage and the docs.
The audit log's search box reported a match count against a total it never searched. The container, action and date filters all query the server, but the free-text box filters only the rows already fetched, because the audit API has no full-text parameter. The filter bar compared that page-scoped match count against the server's full unfiltered total, rendering as 2/500, which reads as "2 of your 500 entries matched" when it meant "2 of the 50 on this page, and nothing looked at the other 450". A user searching for an older event got a near-empty result that positively asserted the event did not exist. While a search is active the count is now against the page and labelled as such. Audit is the only server-paginated list view, so no other view carried this.
The servers detail panel ignored its own Refresh button. Refetching rebuilds every entry, and the selected server was captured at click time and never re-synced, so the panel kept showing the container counts, status and last-seen from whenever the row was clicked while the table row behind it updated. Clicking the panel's own Refresh made that worse rather than better, since it triggered exactly the refetch that orphaned the panel's copy. The agents and containers views already do this re-lookup, so this was a missed site rather than a missing pattern. The panel now re-resolves the selected server by id after every refetch and closes if it is gone.
The self-update helper could destroy a health-verified replacement when removing the old controller failed. The main Docker update path already treated old-container cleanup as best effort after its health gate, but the helper still sent a 409, timeout, or other removal failure into rollback. That rollback force-removed the healthy replacement first and could then fail to restore an old container that Docker had already reaped. The helper now treats a missing old container as already cleaned up and records every other cleanup failure on the successful operation without rolling back the replacement.
A container seen before its registry was configured stayed stamped unknown forever. shouldRepairStoredImageReference() only re-derived a stored image reference when the tag was unknown or digest-shaped, so a container whose image.registry.name was already unknown but whose tag was otherwise normal never re-entered the repair path on refresh, and only recovered if it was recreated. The repair now also runs when the stored registry name is unknown, re-resolving it from the live image inspect on the next refresh cycle. Reported by @depuits in #945.
DD_AGENT_ALLOW_INSECURE_SECRET was parsed as an agent named allow. getAgentConfigurations() handed the whole ddEnvVars map to the generic dd.agent prefix parser, and that documented flat flag matches the prefix as DD_AGENT_ALLOW_INSECURE_SECRET, producing an {insecure: {secret: 'true'}} agent literally named allow. Registration then rejected it every boot with Agent allow failed to register ("host" is required), on every install that set the flag regardless of whether any real agents were configured. The flag is now excluded before the map is parsed. Reported by @depuits in #945.
Debug dumps no longer expose Apprise service URLs, Rocket.Chat user IDs, or Telegram chat IDs. These provider-specific credential fields are now redacted without hiding ordinary parse.urls configuration flags.

1.7.0-rc.7 — 2026-08-29

Security

Every cookie-less request authenticating with an Authorization: Basic header persisted a 30-day session row that was never reaped. requireAuthentication called passport.authenticate(getAllIds(), { session: true }) unconditionally, and passport's req.logIn regenerates and saves a session regardless of saveUninitialized: false, which only governs the auto-save on response end. Cookie-less Basic polling against /auth/user is the documented pattern for wud-card and Homepage integrations, so a single poller accumulated one session document per request. Worse, connect-loki never receives a ttl (only ttlInterval), so LokiJS's own expiry sweep never ran and the store grew /store/dd.json without bound until it was rebuilt by hand. requireAuthentication now passes session: false whenever the request carries an Authorization header, so passport still authenticates the request but skips the regenerate/save entirely; cookie-based logins through /auth/login are unaffected.
A container could point another service's compose image: at its own repository. The compose action picks which service to rewrite from the container's own com.docker.compose.service label and never compares the two images, so a container carrying a label for a service it is not an instance of had that service's image: line rewritten to its own repository, and the operator's next docker compose up would run that image with the victim service's volumes and privileges. The one runtime-versus-compose comparison there was, reconciliationMode, compares full references and so cannot tell a tag drift from a different repository, and in its default warn mode it only logs. The write also lands before the runtime update and is rolled back only if that update throws, which it doesn't, because it is the labelling container's own recreate. Setting a label needs write access to the socket, which is root on plain Docker, so this is only a boundary in the shapes where container-create isn't: Portainer with bind mounts disabled for non-admins, rootless Docker with one watcher per user socket, an authz plugin scoping mounts per client identity. A repository mismatch between the compose image: and what the container actually runs now throws before anything is written, in every reconciliationMode including off, because a different repository doesn't mean the file has drifted, it means the container is not that service. The check runs again under the selected compose file's lock against a freshly read effective multi-file chain, so concurrent external edits and images inherited from another chain file cannot bypass it; automatic backups are created only after that validation passes. Tag-only drift keeps its existing warn, block and off behaviour, and the Docker Hub aliases (docker.io/, index.docker.io/, registry-1.docker.io/, library/) plus digest-pinned references still compare equal to the same repository (#938).
The HTTP trigger's bearer credential was written to the log in cleartext on every registration. Component.register() logs the configuration it was handed, and both layers that are supposed to scrub it missed this one field: the shared trigger redactor knew token, password and apikey but not bearer, and Http had no maskConfiguration() of its own, so the base implementation handed the configuration back untouched. The credential sits nested under auth, which is also why the flat maskFields() helper the other providers use would not have reached it. That log line is not only container stdout: it feeds the buffered stream behind the log API, so the credential was readable over HTTP as well as on disk. bearer is now an infrastructure key in the shared redactor, which covers both the registration log and the /api/v1/triggers response, and Http masks the nested bearer and password directly so the mask holds for any caller that reads maskConfiguration().
GET /api/v1/debug/dump returned cloud registry credentials, chat bot tokens and agent secrets in the clear. The dump's redactor split a key on non-alphanumerics and required a whole segment to equal one of its known tokens, so every compound single-word field name slipped past it: SECRETACCESSKEY, ACCESSKEYID, CLIENTSECRET, BOTTOKEN, ACCESSTOKEN and AGENTSECRET all came back verbatim, for ECR, ACR, Telegram, Matrix and the agent secret. The providers' own maskConfiguration() masks those fields correctly and state.registries/state.triggers go through it; the leak was the separate environment.ddEnvVars block, which is a flat copy of every DD_* variable with __FILE secrets already resolved to plaintext. The strong tokens now match anywhere in the key, while the short ones (pass, key, pat) stay segment-exact so COMPASS_MODE, KEYFILE and DISPATCH are still readable. Pushover's user is its user key rather than an address, so it is resolved by provider rather than by widening user for everyone and hiding the SMTP mailbox with it. Any value carrying URL credentials is redacted regardless of its key, matching what GET /containers already did. This is the fourth fix to the same class of bug, which is why the docs now state the actual rule instead of "all sensitive values are redacted".
The debug dump served raw container environment variables, and no audit or rate limit sat on the route. It read containers through getContainersRaw(), whose own comment says it is "for internal callers that do not return container data to users" — but the dump is downloaded and pasted into support threads. POST /:id/env/reveal shows the same values behind a 10-per-minute limit and an env-reveal audit row; the dump had neither, and it also missed the substring and URL-credential checks that path applies, so MINIO_SECRETKEY and REDIS_AUTHTOKEN were readable through it and not through the API. The dump now reads the same redacted clones GET /containers serves, records a debug-dump audit entry on success, and is limited to 5 requests per minute.
Six registries matched lookalike hostnames and sent the operator's credentials to them. ACR, GHCR, GCR, Quay, LSCR and DHI decided whether an image belonged to them with a regex of the shape /^.*\.?azurecr.io$/ — an optional separator, and in four of the six an unescaped dot — so evilazurecr.io, evilghcr.io, ghcrXio and ghcr.io.attacker.com all matched. Provider selection prefers a credentialed instance, and the request keeps the image's own host, so a container simply running evilghcr.io/victim-org/private:1 (a typo-squat, or a compromised compose file) was enough: no label needed. ACR then attaches a static clientid:clientsecret Basic header unconditionally; the others mint a pull token from the operator's PAT at the real token endpoint and send it to the lookalike host. All six now use a shared matchRegistryHost helper that accepts the base host exactly or as a dot-suffixed subdomain, which is what Hub, DOCR, Mau and Trueforge already did — the loose form was inherited from the upstream project's first commit, not chosen. Regional GCR (eu.gcr.io, asia.gcr.io) and per-tenant ACR (myregistry.azurecr.io) are unaffected. matchUrlPattern had no callers left and is gone.
The command trigger's env sanitizer stopped shell injection but not argument injection. It replaced the metacharacters `, $, ;, &, |, <, >, ( and ) with _, and left spaces, -, and the globbing characters alone. The docs' own canonical example expands a variable unquoted, so image_name set to --registry evil.example/backdoor * reached an operator's printf "[%s]\n" $image_name as --registry, then evil.example/backdoor, then one word per file in the working directory — a flag and extra arguments injected into whatever the script does with them. Reachable without touching a container: POST /api/v1/triggers/command/:name documents caller-supplied container JSON as simulated test data, so any authenticated caller could set the field. Scalar container fields now also lose whitespace, *, ?, [, ], {, }, ~, \, " and ' to _, and a leading - is dropped rather than replaced so the value cannot look like an option at all. container_json, containers_json, dd_title and dd_body are exempt and keep their spaces and quotes, because the JSON has to stay pipeable into jq and the digest title and body are message text; the docs now say to double-quote every expansion and the examples do. The lifecycle hook sanitizer, which its own comment says matches this one, moved with it — hook values are all scalars, so nothing there legitimately carries a space. One visible consequence: a dd.display.name with spaces now reaches a command script with underscores in their place.
An agent could still change how the controller pruned that agent's own containers, by naming an id it doesn't own. The ownership gate added for the bulk ingestion paths lives in processAuthoritativeContainer, and all four callers prune before they ingest, so a foreign id in a sync frame still reached pruneOldContainers unchecked. It could not delete or save a record on its own, because the prune only ever considers rows already owned by the reporting agent. What it could do is the #496 replacement match, which is keyed on agent::watcher::name with the agent forced to the reporting agent: naming a container owned by another agent or by the controller was enough to turn the removal of one of the reporting agent's own rows into a replacement, which retains the update policy on the incoming record and skips the Home Assistant discovery cleanup for the one going away. The prune now filters its input through the same rule the ingest gate uses, silently, since the ingest pass logs each rejection a moment later.

Fixed

A self-update could restart the controller while an unrelated container update was between removing the old container and starting its replacement. Self-update bypassed the optional global concurrency cap and shared no lock with a container in another Compose project, so both lifecycles could start together. The new controller then recovered the unrelated operation against its deleted container id, got a 404, and left the replacement in Created. Self-update now takes a fair process-wide exclusive lifecycle gate: it waits for active updates to finish, later updates queue behind it, and a successful helper handoff keeps the gate closed until the process restarts. A failed or dry-run handoff releases the gate, and infrastructure-mode updates keep their existing global-cap bypass without becoming permanently exclusive. Reported with the operation records that exposed the race by @tarzan77cz in #930, fixed in #942.
A caller-supplied update operation id was inserted without checking whether it already existed. POST /api/v1/triggers/:type/:name and its per-agent variant accept an operationId in the body so the controller can thread its own id through to an agent-local route. Nothing checked it, and the store's id index is not unique, so reusing an id left two rows sharing it. Lookups return the newest of the two, so every later status write landed on the duplicate and the original stayed queued, holding its container's active-operation gate for the full 30 minute TTL or until a restart reconciled it, and when the orphaned entry's turn came round it resolved to the other container's row and overwrote it with its own fields. The active-operation gate did not catch this because it only looks up operations for the container being targeted, so an id belonging to a different container passed straight through. Both routes now return 409 for an id that already exists, active or terminal, and nothing is inserted.
Self-update stranded itself on hosts that need the root break-glass pair, leaving nothing running under the container's real name. The transition helper is the drydock image with only its command overridden, so it still runs the entrypoint as uid 0, but its environment was built from scratch and carried only the DD_SELF_UPDATE_* values. On a host where the Docker socket is owned by GID 0 — Docker Desktop, OrbStack, a root-owned socket, the population the FAQ tells to set DD_RUN_AS_ROOT and DD_ALLOW_INSECURE_ROOT — the entrypoint refused implicit root mode and exited 1, and AutoRemove deleted the evidence. Nothing noticed: dockerode's start() resolves when the container execs, not when it exits, so the code logged "Helper container started" and reported success while the old container sat renamed to -old-<timestamp>, the new one had never been started, and the next attempt failed on the rollback cascade guard. The helper now inherits both variables from the running container's own inspected environment when both are true (both, because either one alone still gets refused), and a watchdog inspects the helper a few poll intervals after starting it: if it has already exited, or auto-remove has already reaped it, the rename is rolled back and the failure is reported with the exit code instead of being swallowed. Reading the pair from the container spec rather than process.env keeps it working for an agent-side spawn.
An update that had already passed its health gate could still be rolled back, and with AutoRemove set that could leave nothing running at all. The Docker update strategy renames the old container out of the way, creates and starts the new one, gates on health, and only then removes the old one. That last removal rethrew anything that was not a "no such container", which sent an already-finished update into the rollback path: the health-verified new container was stopped and force-removed, and the old one renamed back and restarted. The errors that reach it are ordinary Docker hiccups, not failures of the update — the 10 second timeout on waiting for an AutoRemove container to disappear, a 409 removal of container <id> is already in progress, a 500 from a slow overlay unmount or an anonymous-volume delete under IO pressure. When the old container had AutoRemove set the outcome was worse than a rollback, because Docker may already have deleted it by the time the rename back ran, so the rollback failed too and the user was left with no container at all. Cleanup after the health gate can no longer roll anything back: the new container stays running, the operation is marked succeeded, and the leftover <name>-old-<timestamp> container is logged as a warning and recorded on the operation so it can be pruned. The two tests covering this pinned the old behaviour — they were written for branch coverage rather than as an assertion about what should happen — and now assert the update survives.
Every sanitized log line kept the visible half of an ANSI escape. sanitizeLogParam stripped control characters first, which deletes the escape byte the ANSI pattern then looks for, so the pattern could never match and hello\x1b[31m world came out as hello[31m world instead of hello world. Not an injection hole, since the escape byte itself was always removed and a terminal can't act on the remainder, but the module's own contract said it removed ANSI sequences and it didn't, the residue is in every log line, audit detail and close reason that carries user- or registry-supplied text, and the test asserting the behaviour had the residue written into its expected value. ANSI now goes first, and the pattern covers both ESC-prefixed and C1 CSI sequences across the whole CSI grammar rather than just color sequences ending in m, so a cursor or clear-screen sequence doesn't leave [2J behind either; the remaining C1 control range is stripped with the other controls. The scanner-asset error path had grown its own local copy of the ANSI regex to work around this, with a comment explaining the ordering problem; that copy is gone (#938).
Any repository with more than 1000 tags was silently truncated at the first page, and on AWS ECR Public it failed the watch outright. Tag listing captured the Link header the registry returned and then discarded it, rebuilding the next-page cursor by hand as last=<the previous page's last tag>. Pagination cursors are opaque under the OCI distribution spec, so that only works on registries that happen to accept a literal tag name there. ECR Public does not: it answers the rebuilt cursor with 405 Method Not Allowed and Invalid parameter at 'NextToken', which is not in the retryable set and so surfaced as Error when processing (Request failed with status code 405) for that container, every cycle. Found by running the published rc.6 image against a real Docker host: public.ecr.aws/supabase/postgres has 1502 tags, page one returns 1000, and the installed tag existed only on the page that never loaded. Sibling images on the same registry were unaffected because none of them cross 1000 tags, which is why this reads as a per-image fault rather than a registry one. The cursor is now followed as given. A cursor pointing at a different origin is refused rather than followed, because the request carries registry credentials, and tag listing stops after 50 pages so a registry that always reports another page cannot loop forever. Coverage was 100% on this code the whole time: the existing tests drove the pagination loop with link: 'next' and link: 'rel="next"', neither of which is a Link header, so the loop condition saw a truthy value and the cursor-building path was never given anything real to parse (#927).
AWS ECR Public was rate-limited at the generic default and drew 429s from ordinary use. The per-host token bucket has tuned entries for GHCR, Docker Hub and the GitHub API, but ECR Public fell through to the more permissive 5 requests per second default despite throttling anonymous reads harder than any of them. A compose stack pulled from a single ECR namespace, Supabase self-host being the common case, puts a dozen containers on that host in one cycle and produced widespread 429 warnings. It now uses the same 2 per second and burst of 10 the other strict registries get.

Performance

Turning security scanning on could disconnect every open browser tab once per scan cycle. The dd:container-added and dd:container-updated server-sent events carried the whole container record, which includes the per-CVE arrays for both the running image and the update candidate (capped at 1000 entries each, around 340 bytes per entry), the SBOM documents and the cosign verification blocks. That is roughly 340 KB for a single container against the 256 KB a client is allowed to have pending while its socket is backed up, so a scan cycle touching two or three replicas of a vulnerable image dropped every listener, and the reconnect replayed the same oversized events out of the ring buffer and dropped them again. The container list endpoint has stripped those fields since it was written; the event path never did, and there was no projection between the store and the wire. Both lifecycle events, and the removal event, now go through that same projection, on the wire and in the replay ring, so they carry exactly what the list carries: scan status, severity summary, blocking count, block severities and scan time. No display loses anything, because the UI never read the arrays off the event — vulnerabilities, SBOMs and signatures are fetched from their own endpoints, and the security detail is refetched on dd:scan-completed regardless.

Documentation

The public feature comparison was dated March 2026 and measured drydock against a field that no longer reflects what people are choosing. One table ran against WUD, Diun and two archived projects, so the three tools that draw the direct comparisons today, Arcane, Komodo and Dockhand, appeared nowhere in it. It is now two tables, update managers and management platforms, so neither gets wide enough to stop being readable, and every cell traces to the 2026-08-29 competitive audit. Corrections that came out of that audit: WUD ships threshold filtering and has 17 notification providers and 12 registries, not 16 and 13; Diun's semver and Home Assistant support are partial rather than full, and it does have Prometheus metrics; Watchtower's Shoutrrr count is about 20. Drydock's own rows now include the two it loses, RBAC and a real pending-approval queue, rather than only the ones it wins. A note under the first table records Watchtower's December 2025 archival, its v1.7.1 (November 2023) last release, and the unofficial community fork still shipping releases. The same table is mirrored into all six translated READMEs.
Seventeen rows across the comparison surfaces on the site disagreed with the audit, and two of them undersold drydock or a competitor. Komodo ships maintenance windows, and Dockhand ships Prometheus metrics and Home Assistant MQTT, all three of which were listed as absent. Komodo and Dockhand both have a limited dry-run rather than none, and Dockhand has no rollback on failure where the /compare index called it partial. Diun's semver and MQTT rows claimed parity where the audit found partial support. Dockge's stable releases have been stalled since March 2025, so it is no longer listed as plainly maintained, and it ships 29 UI languages, not "30+". Portainer does have a distributed agent architecture, which the homepage teaser denied. And the Dozzle page still called drydock's resource monitoring planned when it shipped in v1.5.0. The WUD and Watchtower provider counts now agree with the README, and Drydock's 23 registries are 11 more than WUD's 12, not 10 (#938).
The README roadmap table listed scoped rotatable API keys under v2.0+. They are v1.8.0 work now, where the SQLite store migration provides the persisted key table they depend on, and the per-update approval queue is listed with them.

Chore

A compose rollback that itself fails was mapped to a terminal state no test ever drove. getComposeRollbackTerminalPatch turns a rollback-failed compose outcome into {status: failed, phase: rollback-failed}, but the whole branch sat under a v8 ignore claiming it was integration-covered through compose recovery, and nothing drove that status through runContainerUpdateLifecycle: the one lifecycle test used rolled-back, and the compose-side test stops at the throw. A regression mapping it to rolled-back, or letting it fall through to the duplicate-update expired reclassification, would have shipped at 100% coverage. The second ignore in that path, over the rollback-state persistence, was stale the other way: the rolled-back lifecycle test has been exercising it all along. Seven new lifecycle cases now cover both statuses plus the blank-reason, absent-lastError, unrecognized-status, no-container-id, no-persistence-dependency and watcher-only-identity paths, which retires seven of the file's thirteen v8 ignore blocks. The terminal patch's lastError is also typed as always present, which it always was, so its dead fallback is gone. The six ignores left each name a specific unreachable input instead of claiming coverage that lives somewhere else.

1.7.0-rc.6 — 2026-08-29

Added

Security

An authenticated agent could still pre-empt a controller-local container before the controller wrote it. The ownership check landed for #904 only covers a container id the store already holds; a brand-new id had no check at all, so an agent could insert a record under a watcher name matching the controller's own local Docker watcher, race the controller's own watch cycle, and have docker-trigger.ts's trigger routing (which selects purely on the stored agent field) hand that container's future lifecycle actions to the agent instead of the controller. A no-record insert claiming a watcher name the controller registered directly, rather than through an agent, is now rejected.
That #904 fix only covered the two incremental dd:container-* event handlers, not the bulk ingestion paths. Handshake, the watcher-snapshot fallback, on-demand watch/watchContainer, and edge handleContainerSync all reach buildContainerReport through processAuthoritativeContainer with no ownership check in between, so an agent could still name an id owned by another agent or by the controller's own local watcher in its next routine snapshot and have the store reassign it, repeatedly, at its own cadence. processAuthoritativeContainer now runs the same no-existing-record and different-owner checks before ingesting. It deliberately skips the third check canMutateContainer already has — rejecting a watcher mismatch against the stored record — because bulk ingestion is exactly how a legitimate watcher rename propagates (an operator renaming a DD_WATCHER_<NAME>_SOCKET key), and rejecting on that mismatch here would silently stop every future report for that container id from landing, with nothing left to prune or retry it.
Thirteen registries authenticated for the version check and then pulled anonymously. The pull-credential builder handled a login and password pair and a username and token pair, but had no branch for a configured auth value, while the sibling that builds lookup credentials did. So Hub, Custom, DHI, DOCR, Harbor, Gitea, Forgejo, Codeberg, Nexus, Artifactory, Alibaba CR, OCIR and IBM CR detected an update correctly and then attempted the pull with no credentials, which fails outright against a private registry and consumes the anonymous quota against a rate-limited public one. The value is decoded into a credential pair rather than passed through, because the Docker daemon reads the pair and never inspects an auth sub-field, and it is split on the first colon only since a password may contain one. A malformed value now fails closed rather than returning nothing, which would have silently reproduced the same anonymous pull for a different bad input.
A watcher failure on an agent returned the thrown message to the caller. Two agent handlers returned the raw message for anything that was not an Error, which is the branch a rejected upstream request lands in with its context attached. Both now return the fixed generic message unconditionally; the detail still reaches the server-side log through the existing sanitizer.
A bulk security scan request had no cap on how many containers it could queue. parseBulkScanBody validated that containerIds was an array of non-empty strings but accepted it at any length with duplicates intact, and each accepted id drives a vulnerability scan in the handler, so an authenticated caller could send tens of thousands of ids, or the same id repeated, and make the controller do that much scanning work from one request. The sibling parseWatchContainersBody already caps and dedupes for the same reason; bulk scan now enforces the same 200-entry limit, checked before the per-element loop so an oversized array is rejected without walking it, and collapses duplicates with the same Set-based dedup that silently drops a repeat instead of erroring.
Error responses could carry upstream credentials back to the caller. Eight handlers across the trigger, container and agent APIs interpolated a raw thrown message straight into the response body. That matters because of what those messages contain: a trigger failing against an upstream service throws with the request context attached, so an Authorization: Bearer header and a credentialed webhook URL both reached the client in a 500 that any authenticated caller could read. All eight now route through the sanitizePreviewErrorReason scrubber that already existed for exactly this and had only ever been wired into the preview path. The two sites that deliberately tell an empty message apart from a missing one keep that distinction and scrub only a truthy string, so their synthesized fallback still fires.
The preview-error sanitizer missed credentials embedded in a URL path segment, not just in headers or userinfo. Telegram builds its API URL as https://api.telegram.org/bot<credential>/sendMessage and IFTTT as https://maker.ifttt.com/trigger/<event>/with/key/<credential>, so a bot credential or Maker key sits in the path itself rather than behind user:pass@ or an Authorization header. This is a latent gap, not an observed leak: nothing currently puts either provider's request URL into an error.message, but a future throw that includes request context would carry the credential straight through. sanitizePreviewErrorReason now redacts the credential segment for both, plus the same shape in Discord webhook URLs (https://discord.com/api/webhooks/<id>/<credential>) found while checking the other providers, all anchored to each provider's own fixed host and path rather than to "any long opaque string" so an ordinary registry path segment, including a manifest digest, is left alone.
The API end-to-end suite could not detect an unmasked secret. Five assertions in api-registry.feature checked masked credential fields against .*, which passes for any value including a completely unredacted one. Component.mask() returns the literal [REDACTED], so all five now assert that, matching what the authentications feature already did.

Fixed

Row selection never highlighted anything in seven views. Servers, Registries, Watchers, Audit, Auth, Notifications and Triggers all passed an active-row prop to the shared data table, which declares selectedKey. Vue drops an undeclared prop onto the root element and carries on, so the selection styling was a no-op in every one of them. The test suite was fine with it because each view's table double accepted the invented prop and echoed it back into an attribute the spec then asserted against, so the specs were checking the stub's own invention rather than the real component's contract. Coverage was 100% throughout. The doubles now require the real prop and each row-click test asserts the value actually changes.
White text on a light accent, as low as 1.37:1. The trigger test button, the config profile avatar and the app-layout user menu avatar hardcoded white over a gradient whose far stop is the info or success color. Flattening to the solid primary token with its computed foreground clears 4.5:1 in all twelve themes, worst case 5.07:1. Keeping the gradient could not be fixed with the correct token, because the foreground token is computed against the primary color and not against the other stop. The contrast gate now also scans the sources for a tag combining white text with a primary-color reference, since token math cannot see inside a template, which is how this survived the previous contrast pass.
The notification outbox showed whichever tab's data resolved last rather than whichever tab was selected last, so switching quickly left the other tab's rows on screen. Guarded with the request-id pattern the triggers and notifications views already use.
Status was unreadable or untranslated in five places. The icon-only status indicators in Servers, Registries and Watchers had no accessible name, so state was conveyed by color and glyph alone. Auth, Audit, the container log header and the dashboard security widget interpolated raw enum values, so a non-English reader saw active, success, error and CRITICAL in English regardless of locale. Five new English keys, translated into all 16 locales in the same change rather than left for a sync to fill with verbatim English.
Removing a container while its full-page detail view was open left a blank screen. Two independent causes, either of which still reproduces alone. Closing the panel nulled the selection without clearing the full-page flag, so the layout stayed in full-page mode with nothing to render. And the watcher meant to notice the removal used a getter source: a watch with a getter source fires on reference change, not on in-place mutation, and the SSE pipeline removes by splice on purpose to avoid per-row invalidation, so the sync never ran and the selection kept pointing at a deleted object.
Two dashboard watchers missed every in-place SSE update. The pending-update rows and the operation holds both watched a bare ref, so neither ran while the patch pipeline mutated rows in place, leaving ghost update rows and holds that never released. The first takes a length-augmented source; the second watches a per-row fingerprint, because the fields it reads are patched without touching array length. Neither uses a deep watch, which would re-fire on every unrelated field mutation of every row and defeat the point of patching in place.
A malformed auth registry value could still decode into credentials that looked valid. decodeAuthCredentials fed the configured value straight to Buffer.from(auth, 'base64'), which silently drops characters outside the base64 alphabet instead of throwing: dXNlcjpwYXNz! still decodes to user:pass, the trailing ! just discarded. That defeats the malformed-auth throw this decode exists to enforce, since a corrupted or truncated value can still land on a valid-looking login:password split and fall through as real credentials instead of being rejected. The value is now re-encoded and compared against the original before decoding; any character the decoder ignored makes the round trip diverge and the throw fires.
A rare upstream build race could fail a release cut with nothing to recover from. An open BuildKit bug (moby/buildkit#7089) makes a canceled-then-retried operation inside a single multi-architecture build get the QEMU emulator path prepended twice, so the build dies instantly with Invalid ELF image for this architecture before producing any output. It fired once in CI and has passed on every run since, which is what a scheduler race looks like. The multi-architecture smoke build now retries, and the release cut gains a full-build retry for the specific case where the first attempt produced no digest at all. That case matters because the existing manifest retry needs a digest already sitting in a registry and hard errors without one, so before this the race firing during a real cut meant a hard failure discovered live in the release window.
A long enough error killed the process instead of closing a log stream. The container log stream used the raw thrown message as the WebSocket close reason. The library throws synchronously once that exceeds 123 bytes and never catches it, and the stream is invoked as a floating promise with no rejection handler, so a registry error past the limit took the whole process down. Both remaining sites now truncate through the helper the file already had and one sibling was already using. The second of them was never a crash: it sat inside a try/catch and so swallowed the error, and because the socket is marked closing before the throw, the close frame was never sent and the socket stayed open.
Checking for updates again reset the image-maturity clock. Four agent entry points processed the authoritative container list and then pruned, so a container that had just been re-inserted was compared against a store that still held it. Pruning now runs first at all four. The watch path was also missing the zero-container guard the handshake path has, and needs it, because the Docker watcher returns an empty report array both when there genuinely are no containers and when enumeration threw and was swallowed, and the response cannot tell the two apart, so a transient socket error on an agent pruned every container for that watcher. (#565)
The docs search returned every archived version at once. A search for a common term returned about 1600 hits spread across five documentation versions, ranked with the oldest changelog first, so a reader on the current docs was served entries describing a configuration prefix this release removes. Each page is now indexed with its version and the search dialog scopes to the version being read, defaulting to the current one elsewhere.
Two badges on the homepage were blocked by the site's own content-security policy. The best-practices and coverage image hosts were missing from img-src, so both rendered broken and logged a violation in any enforcing browser. Exactly those two origins are added, and the policy's full source list is now pinned by a test.
The weekly DAST scan has never completed. The active scan alone consumed 39 minutes 46 seconds of the job's 40 minute budget, so the job was cancelled the moment the second scanner's steps opened: the second scanner has never run, the report file was never written, and the severity gate and SARIF upload both ran against a file that did not exist. The two scanners now run as separate parallel jobs with their own budgets, so neither can starve the other and a blown budget names the scanner responsible.
The action-filter documentation described shipped behaviour as scheduled. The callout still framed the include and exclude filters becoming hard blockers as future work, three release candidates after it shipped, so an operator reading it concluded they still had time to migrate labels while their update button was already being refused.
The PUT /api/settings deprecation pointed consumers at a dead endpoint. The deprecation warning logged on every call, and the matching entry in deprecations, both named the unversioned /api/settings path. v1.6.0 removed the unversioned /api alias and settings is not one of the four wud-card compatibility endpoints exempted from that, so anyone following the migration advice moved off a deprecated method and onto a path that returns 410. Both now name /api/v1/settings, and the deprecations entry says outright that the unversioned path is not an alias for it.
The UI is properly translated in all 16 non-English locales. Every locale already had every key, but between 17% and 29% of the values in each were still the English source text showing through, so large parts of the container list, the update and rollback dialogs, the search palette and the notification outbox rendered in English regardless of the language selected. 2109 strings are now actually translated. The 61 left alone are the ones with nothing to translate, such as "{registryHost} — {error}", and category labels whose only word is a product name each locale already keeps in English.
The image-maturity badge showed no age outside English. Detected {duration} ago had replaced an older bare NEW label in the English source, but all 16 locales still carried a translation of the old wording, so the badge silently dropped its duration for every non-English user. Retranslated in all 16, and a new placeholder-parity gate now fails the build when a translated string loses or invents an interpolation placeholder, which is the class of bug that hid this one: it breaks neither JSON parsing nor key parity.
A non-numeric tail or since on the log and agent endpoints put NaN into the ring-buffer read instead of rejecting it. Both now validate and return 400, matching what the audit endpoint already did. No clamping, because the log buffer caps at its own maximum regardless of what is asked for, so an oversized value cannot cause unbounded work the way the audit limit could against a growing collection.
An empty tail or since (?tail=) was read as absent rather than invalid, on all three log endpoints. Every one of them guarded with if (!value) return undefined, and an empty string is falsy along with null and undefined, so a supplied-but-empty value skipped the 400 above it and was silently forwarded downstream as undefined instead of being rejected as malformed input. The controller's log and agent endpoints and the agent-mode API's own log endpoint all now treat only a genuinely missing parameter as absent; an empty string proceeds to the integer-shape check like any other supplied value and is rejected.
A limit or offset on the audit endpoint with a numeric prefix, such as ?limit=25logs, validated as the leading digits instead of being rejected. Number.parseInt stops at the first non-digit rather than requiring the whole string to be a clean integer, the same bug already fixed on the log and agent endpoints' tail/since. Audit's contract falls back to its default instead of erroring on a bad value, so that stays: an unparseable limit/offset now behaves the same as an absent one and falls back to the default, rather than switching audit over to a 400 like the other three params. A value that overflows Number.MAX_SAFE_INTEGER gets the same treatment, closing a related precision-loss gap where a huge digit string used to silently clamp to 200.
A duplicate Portwing agent key was reported as a server error. The conflict now returns 409 with the conflict message, carried by a typed error rather than by matching on the message text, which is how the update path already does it.
The Dutch curl healthcheck banner named the wrong thing. Its title rendered the CLI tool curl as krulstatus, which in Dutch is a decorative curl.
Drydock offered older LinuxServer builds as updates. v8.9.0-ls101 was reported as having an update available to v8.9.0-ls99, and the same for 0.5.0b3.dev101-ls250 down to 0.5.0b3.dev99-ls235 and 4.0.19.2979-ls322 down to 4.0.9.2244-ls257. Three separate mechanisms, all landing in the same comparator. The -lsNNN cases were spec-correct semver behaving wrongly for the shape: an alphanumeric prerelease identifier compares lexically, so "ls101" < "ls99" on the first differing character. The four-component case was worse and not truncation: semver.clean('4.0.19.2979-ls322', {loose: true}) mis-splits the 19 and returns the corrupted 4.0.1-9.2979-ls322, while the same call on 4.0.9.2244-ls257 returns null and falls through to coerce() for a clean 4.0.9, so the two sides were never compared through the same parse path at all. Candidates whose version cores tie are now compared on the numeric counters embedded in an otherwise identical suffix template, generalised to any trailing counter rather than special-cased to LinuxServer, and a pair whose ordering cannot be established safely is declined rather than guessed. Note this was visible on default settings too: DD_WATCHER_DOCKER_TAG_FAMILY=loose was needed for the downgrade to be offered as actionable, but the informational "newer version available" insight shared the same admission logic and showed the older tag without it. (#918)
The action-policy badge and its tooltips rendered in English in every locale. The Auto/Manual/Blocked badge and the three state tooltips behind it, on both the update-status panel and the per-trigger rows, were added to the English catalog only, so eleven strings fell back to English for every non-English user from the release that introduced them. Translated in all 16 locales. The key-parity gate now fails on a key that exists in English and not in a locale, which is what should have caught this: it previously tolerated the gap on the grounds that Crowdin would fill it after the push, and what Crowdin actually fills it with is verbatim English.
The weekly Crowdin sync no longer reverts the six translated READMEs to English. README.md was registered as a Crowdin source, so every sync round-tripped the translated copies through the project. Turning off content segmentation in #804 to fix an earlier corruption renumbered every segment, which orphaned the stored translations from the strings they belonged to, and Crowdin exports source text for anything it cannot match. The next sync opened as a pull request deleting 1392 lines of translated README, caught only by readme-translations.test.ts. Crowdin no longer maps README.md at all: the translated READMEs are hand-authored in-repo and asserted phrase by phrase at every cut, and nobody was translating them in Crowdin. The UI locale catalogs are unaffected, since segmentation never applied to JSON. (#919)
Restoring the maturity-policy stash at startup no longer ignores the size cap, or evicts the wrong entry afterwards. The durable updatePolicyRetentionCache added in 1.7.0-rc.2 restored every non-expired persisted record straight into memory without checking DD_UPDATE_POLICY_RETENTION_CACHE_MAX_ENTRIES, so lowering that limit between restarts came back over it and the extra entries stayed live until the next stash trimmed them. Restoring also used the store's document order, which has nothing to do with stash age, and the eviction that runs when the cap is hit reads that order as its LRU, so the first eviction after any restart could drop the newest entry and keep the oldest. Startup now restores oldest first and enforces the cap against both the in-memory cache and the persisted records. (#565)

1.7.0-rc.5 — 2026-08-27

Security

An unauthenticated Portwing hello could throw outside the callback error boundary. A non-string compatibility value in the hello payload reached .split() before anything validated it. The payload is now validated before parsing. (#904)
An authenticated agent could write to containers it does not own. Update and removal accepted globally keyed container IDs belonging to another agent or to the controller; ownership is now enforced at the ingestion and removal boundaries. (#904)
Runtime environment redaction missed *_PAT values and credentials embedded in URLs. Both are now detected and redacted in the debug dump. (#904)
Rejected-origin diagnostics could be used to amplify debug logging outside the route limiter. The diagnostic path is bounded and rate-limited. (#904)
Credentials survived redaction in scheme-relative container URLs. A value like //user:pass@host/path has no scheme to anchor it, so parsing threw and the raw string was returned unredacted. Those values now parse and redact like any other URL. (#904)

Accessibility

Dark themes now meet the WCAG 2.2 contrast minimum for normal text. Secondary and muted text, the tone colors (warning, caution, danger, success, info) and the two accent colors all failed 4.5:1 against surfaces they are actually painted on, across all six dark themes. Worst measured pairs were muted text on an elevated surface at 1.57:1 and secondary text in a warning dialog at 1.33:1. Token values are raised to clear 4.5:1 against the worst real pairing for each, keeping each palette's hue. Colors used only for borders, focus rings, icons and toggle fills are held to the 3:1 non-text target instead of being raised needlessly. (#850, #865)
Text in the update-status condition rows is readable again. Each nested condition row painted the same muted tone background its parent section had already painted, compounding two tints into a surface no text color could clear without washing out the whole muted/secondary hierarchy. The rows now use the card surface they already carried, which no value change could have fixed. (#850)
Toast messages meet the contrast minimum. Toast surfaces mixed their tone color at 25%, leaving the tone-colored text on them between 3.65:1 and 3.94:1 in five of the six dark themes. They now use the same 15% mix as every other tone surface in the app. (#850)
Primary button labels are no longer white on a light accent. The login and notification-test buttons hardcoded white text on the primary color, measuring as low as 1.45:1. A new --dd-primary-fg token picks black or white per theme, whichever clears 4.5:1 against that theme's primary. (#850)

Fixed

Startup recovery could mark an untouched container as a successfully updated one. A prepare-phase operation interrupted before Docker did anything was reconciled as though the replacement had happened, so a container that never changed was recorded as updated. Recovery now compares the persisted old and new container identities: an untouched original fails recovery, a real replacement still succeeds. Startup also stopped expiring destructive in-progress Docker updates before anything had inspected the runtime, and recovered dependency groups rehydrate their completed-upstream context instead of running a restart-only dependent as a full update. (#904)
A healthy agent with a large inventory could never reconnect. When the cached watcher replay exceeded 256 KiB the controller rejected the agent before acknowledgement, permanently, so the biggest deployments were the ones that could not come back. The stream now stays open and sends only the acknowledgement, and the authenticated full-inventory handshake supplies the state. Latest-only coalescing also no longer overwrites a protocol-critical snapshot belonging to a different watcher. (#904)
Backups could be selected or pruned across unrelated containers that shared a name. Ownership keyed on the container name alone, so the same service name under two agents, two watchers or two compose projects collided. Backups now carry a stable scoped identity, and legacy identity-less backups stay selectable only while a single active identity owns that name. (#904)
Rollback restored a mutable tag instead of the digest recorded with the backup, so a rollback could land on whatever that tag pointed at by then rather than the image the container was actually running. (#904)
A slow SSE client could make drydock retain unbounded queued bytes. The main broadcast stream, the container and summary stats streams, and the agent controller fan-out all ignored res.write returning false. All three now deliver bounded, drain-aware latest-state updates and clean up on close, and the agent fan-out has a connection cap. (#904)
Concurrent scans could cancel each other. The digest single-flight entry captured only the first caller's cancellation state and transient-retry option, so one caller aborting could kill an unrelated caller's scan, and a later waiter could join an entry whose controller was already aborted. Waiter cancellation is isolated, retry requirements are aggregated across live waiters, and aborted entries are retired before a new waiter is admitted. A cancelled scheduled scan also now propagates cancellation into the scanner instead of rejecting the wrapper and letting later runs accumulate. (#904)
Docker event chunks were parsed concurrently against a shared buffer, racing both the parser state and the writes that followed. Parsing and application are now serialized. (#904)
A successful container action returned 500 when the follow-up refresh failed. The best-effort post-action inspect or store refresh was allowed to overwrite the outcome of an action that had already succeeded. Success is now authoritative and refresh degradation is reported separately. (#904)
Paginated container lists were only sorted within a page. The default name sort ran after the store had already paginated, so ordering was not global across pages. (#904)
A restart suppressed batch-completion events for updates that were still in flight. Volatile batch membership was cleared on startup, so recovered operations completed silently; membership now rehydrates from the persisted operations. (#904)
The system-log stream limiter fell back to an empty identity instead of the client IP when no configured key applied. (#904)
An unsupported agent transport was admitted and then failed asynchronously at watcher lookup during a dependency restart. It is now rejected at admission. (#904)
A backpressured controller SSE client lost every event but the newest. Only one pending payload was retained while the socket drained, so anything that arrived behind it was silently replaced. Pending payloads are now queued in order and flushed on drain, with the byte cap applied across the queue so a genuinely slow client is still dropped. (#904)
Tearing a watcher down while it was setting up its Docker event listener could resurrect it. Deregistration landing during the client-recreation or auth preflight, or during the getEvents request itself, left a late callback to fire against already-cleaned-up state: it registered a stream and attached handlers to a watcher that was being removed, or scheduled a reconnect for one that was already gone. Superseded and deregistered listener setups now bail out at each of those points, and a stream that arrives too late is destroyed rather than adopted. (#904)
An update that never started was reported as a failed update. An operation that could not be resolved before the runtime was reachable was marked failed rather than expired, so users saw a failure for an update that had not run. (#904)
The Crowdin sync workflow failed on every push to a dev branch that was not the newest one. The base resolver always picked the highest dev/vX.Y on origin regardless of which ref triggered the run, but the Crowdin action had already downloaded translations into the working tree by then, so checking out a different branch over them died with Your local changes to the following files would be overwritten by checkout. A push to a dev/vX.Y branch now targets that branch directly; scheduled and manually dispatched runs keep the highest-wins lookup and its default-branch fallback. (run 33047712284)

Documentation

The homepage FAQ described DD_TRIGGER_* as still deprecated, past its actual v1.7.0 removal. It said the prefix was "a deprecated alias scheduled for removal", but v1.7.0 removed it outright: any DD_TRIGGER_* environment variable now fails startup with an error naming every offending variable and its exact DD_ACTION_*/DD_NOTIFICATION_* replacement (see deprecations). The FAQ answer now says so, and a new test asserts the copy can't drift back to describing an already-removed feature as merely pending removal.
Archived changelog snapshots kept relative ./DEPRECATIONS.md links that don't resolve on the docs site. content/docs/v1.x/changelog/index.mdx files are frozen copies of the root CHANGELOG.md, which uses relative links, and the docs sync script's per-version rewrite only handled absolute /docs/... targets. So the relative-link rewrite that already ran for the freshly generated current-version changelog never reached the frozen archives. rewriteDocsLinksInDirectory now routes changelog files through the same rewrite, fixing the two broken links in the v1.5 archive, and any future archive with the same pattern, without touching non-changelog content.

Chore

Two real production modules were excluded from the coverage gate, and a third exclude entry never matched anything. query-values.ts and sorting.ts are used by maturity-filter.ts, container.ts and crud-context.ts, but were carved out of the 100% coverage requirement the repo otherwise enforces. Both are back in: sorting.ts was already fully covered indirectly, and query-values.ts needed a new direct test for the repeated-query-param (array) path, which is real Express qs behavior nothing exercised before. The third entry named Trueforge.ts while the file on disk was trueforge.ts, so it was a silent no-op on case-sensitive CI, and dropping it surfaced what the wrong case was hiding. registerComponent() resolves a provider's module by testing for a Provider.ts-cased file, and on a case-insensitive filesystem that test also matched trueforge.ts, loading it under a second differently-cased specifier the direct unit tests never touch, so full-suite coverage read it as partly uncovered even though the module is fully tested. Every other provider file already follows the PascalCase convention, so the file and its test are renamed to match and the convention check now resolves identically on every filesystem.
Removed the now-stale CVE-2026-14456 (openssl/libssl3/libcrypto3) Grype suppression. It was scoped to the Alpine 3.24 openssl pin at 3.5.7-r0 and named its own removal trigger: bump the Dockerfile's openssl= pin to 3.5.8 or newer. That bump already landed in #881, so openssl is now pinned at 3.5.8-r0, and a live Grype scan of that exact package and version confirms zero matches for this CVE or any other finding. The suppression no longer applies.

1.7.0-rc.4 — 2026-08-26

Security

Base images bumped to clear six HIGH OpenSSL CVEs. The node:24-alpine and alpine:3.24 digest pins are rolled to the current upstream rebuilds, moving the shipped OpenSSL from 3.5.7-r0 (flagged by Grype for CVE-2026-14457, CVE-2026-18798, CVE-2026-54874, CVE-2026-63072, CVE-2026-63075, and CVE-2026-63076) to 3.5.8-r0. (#881)
The demo site sends the full security-header set. demo.getdrydock.com was failing the weekly ZAP baseline scan with six WARN-NEW alerts. apps/demo/vercel.json now sends X-Content-Type-Options, Permissions-Policy, a credentialless Cross-Origin-Embedder-Policy, and a same-origin Access-Control-Allow-Origin; the two cache-control alerts are allowlisted in .zap/rules.tsv under the same static-SPA rationale as the existing 10049 entry. (#878)

Fixed

WebSocket log-stream connections behind a TLS-terminating proxy no longer 403 when X-Forwarded-Proto is absent. With trust proxy enabled, isOriginAllowed fell back to the local socket's encrypted flag whenever the proxy omitted X-Forwarded-Proto, which is plain HTTP on a backend behind TLS termination, so a browser's https:// Origin never matched and every WebSocket upgrade was rejected while REST traffic worked fine. The protocol is now treated as unknown (and skipped from the comparison) in that case instead of being inferred from the local socket; host validation is unaffected. (#867, #868)
A ws/wss value in X-Forwarded-Proto no longer rejects the WebSocket upgrade. Traefik forwards the upgrade's client-facing scheme as wss rather than https (traefik/traefik#6388), which the origin check treated as an unsupported protocol and hard-rejected — so the trust-proxy fix above still 403'd behind a default Traefik setup. ws and wss now map to http:/https: for the Origin comparison; genuinely unknown protocols are still rejected. (#867, #887)
Startup no longer crashes with EPERM when the store volume forbids chmod. The permission tightening added in 1.6.0 now warns and continues on EPERM/EACCES/ENOTSUP instead of throwing, so mounts that reject chmod (NFS/CIFS volumes, non-root containers, some volume drivers) no longer take the whole process down at startup; a genuinely read-only volume (EROFS) still fails fast at startup, because nothing could be persisted there anyway. (#874, #886)
Tag suggestion no longer ranks a bare integer build-number tag above a real dotted version. A bare integer tag (e.g. 168) coerces via semver.coerce() into a fake 168.0.0, which previously outranked a real release like 1.43.3 — for linuxserver/plex, this meant the suggested-tag badge and, with a permissive dd.tag.include filter, the actionable update candidate itself could point at a destructive downgrade. Bare integer tags are now only ever ranked among themselves (never against a real dotted version, and never at all when the population contains any other non-integer version signal, such as a prerelease-only or coercion-lossy tag), sorted numerically rather than lexically. The rule is shared between the suggested-tag badge (tag/suggest.ts) and the actionable non-semver/includeTags recovery path (watchers/providers/docker/tag-candidates.ts) via a new tag/version-population.ts module so the two paths can't drift apart again. (#859, #871)
The debug dump redacted env var names instead of values. Container env vars appear in the dump as { key, value } pairs, and the generic redaction walker matched the pair's key property against the key sensitive-token rule, so a var like HF_TOKEN showed up as "key": "[REDACTED]", "value": "xyz" — the name was hidden and the actual secret was left in plain text. Pair objects are now handled explicitly: the name always stays visible, and the value is redacted only when the name itself matches the sensitive-key rules. (#875, #885)
Containers that drop out of watch scope are now pruned from the store and UI. A container excluded by watchbydefault being off, or by its dd.watch label being removed, previously kept its stale store record forever as long as it still inspected successfully in Docker — inspect success alone was being read as "still tracked" instead of "still in scope," so the only way to clear it was deleting dd.json. Stopped-but-still-watched containers are unaffected and keep their start-button visibility in the UI. (#869, #888)

1.7.0-rc.3 — 2026-08-23

Added

Portwing edge tunnels now carry non-JSON Docker response bodies. A Docker endpoint that answers with a non-JSON body — for example GET /_ping's plain-text OK — previously couldn't cross the edge tunnel intact. The controller's welcome frame now advertises an edge-response-body-b64 capability and decodes base64-negotiated response bodies from agents that support it, falling back to the existing path otherwise. Additive and capability-gated, no protocol-version bump; pairs with the portwing-side change (portwing #206). (#852)

Changed

README badges read live instead of being hand-bumped. The version, license, pull-count, and stars badges in the English README and all six translations now render from live shields.io endpoints (github/v/release with prereleases, github/license, Docker Hub pulls, github/stars) instead of static images. The typed "GHCR 150K+ pulls" figure is replaced by the live Docker Hub count, and the release tooling no longer bumps or validates a static version badge per cut — there is nothing left to drift. (#851)
The Star History chart ships as a light/dark pair and refreshes at the release cut. The README's star chart is now a <picture> block with theme-matched light and dark SVGs, so it follows GitHub's theme toggle instead of the OS preference, and the committed chart regenerates from the release-cut workflow rather than a cron — it can't mutate underneath a tag. (#844, #847)
DAST and workflow-lint gates fail closed. The ZAP scans (baseline and full) no longer pass -I, which had told ZAP to ignore every warning, so findings now actually fail the gate (targeted suppressions belong in .zap/rules.tsv); the pre-push zizmor step errors with an install hint when the binary is missing instead of silently skipping. (#842)
A daily monitor asserts main carries a release tag. Drydock is the first caller of the org's main-is-released invariant monitor: a scheduled read-only workflow that goes red when main's HEAD is untagged, deliberately not a required PR context (a promotion merge is untagged by definition until the cut lands). (#846)

Fixed

The release pipeline no longer trips over its own test infrastructure. The rc.2 cut failed on main's push-triggered CI: a global js-yaml v5 override broke Artillery's load-test jobs (reverted to the v3 range Artillery actually supports, #829), and two Playwright waits were budgeted tighter than the backend operations they cover, so promotion runs raced the app's own state machine (ceilings raised past the app's budgets, fixes #832, #836).

1.7.0-rc.2 — 2026-08-20

Added

Action-policy resolution is surfaced on the API and UI (spec-6.0.1-action-policy.md, slice 5). GET /api/v1/containers and SSE container payloads now include an actionPolicy object (state: blocked/manual/auto, plus triggerId and, when blocked, reason) on updateEligibility, reflecting the winning action trigger's resolved verdict for that container independent of whether it produced a blocker. GET /api/v1/containers/{id}/triggers gains a per-trigger resolvedState field with the same three values, so every candidate action trigger's individual verdict is visible, not just the winner. The container list/detail UI shows a new Auto badge (Update Status panel, full-page Actions tab, side panel trigger rows) wherever a container's or trigger's resolved state is auto, with tooltips explaining the blocked/manual/auto states. New dd.action.auto container label and onauto AUTO mode: under AUTO=onauto, dd.action.include keeps granting manual access but automatic dispatch additionally requires a dd.action.auto match, letting a container opt into manual-only access without picking up automatic execution. Drydock now logs a startup WARN for any action trigger left on AUTO=oninclude that has dd.action.include-matching containers with no corresponding dd.action.auto label (would silently drop to manual-only if switched to onauto), and a second WARN when a trigger is on AUTO=none but a container carries an inert dd.action.auto label for it. Docs: new dd.action.auto label and onauto value are documented on the labels reference, triggers configuration, and update-eligibility pages, including a callout on all three clarifying that AUTO=none/AUTO=all also set the baseline access default (not just automatic execution) — AUTO=none still opens manual access to every container, AUTO=all opens both.
Greptile second-opinion review, summoned behind a label. A new thin caller workflow, .github/workflows/greptile.yml, fires only on pull_request: labeled events for the second-opinion label and calls the org's reusable CodesWhat/.github greptile-summon.yml workflow (pinned to a frozen commit SHA, no floating ref) to post one @greptileai review request per exact PR head. CodeRabbit stays the automatic review lane; Greptile is opt-in only, for security-sensitive diffs, large refactors, or a tiebreaker when CodeRabbit and the author disagree. .coderabbit.yaml gains a labeling_instructions entry so CodeRabbit can auto-apply the second-opinion label itself when those criteria hit, in addition to a human applying it manually. (#751)
release-cut.yml can now cut a maintenance patch from a retired dev branch without touching the cosign signing identity. A new optional source_ref input (e.g. dev/v1.6) builds and releases from that branch's tip instead of main HEAD, for a patch on a line main has already moved past (v1.6.1 after main is on v1.7). The workflow itself still only ever dispatches from refs/heads/main, so the published cosign OIDC identity is unaffected. source_ref is bounded hard — it must be an exact dev/vX.Y branch that exists on origin, share release_tag's line, not be the active (highest) dev line, and pair with a nonzero patch version — and the main-sync drift guard is skipped rather than failed, since there's no drift claim to make against a line main has already superseded. Left empty, every normal cut is unchanged. See RELEASING.md's new "Maintenance cuts" section for the full precondition list, including the manual ci-verify.yml/e2e-playwright.yml dispatch required before cutting.

Changed

Documented the container SBOM attestation in the release-verification guide. release-cut.yml has generated and attested a signed SPDX 2.3 SBOM for every release container image since the v1.5 hardening batch, but content/docs/current/guides/verifying-releases never told readers how to check it. Added a "Verify the container SBOM attestation" section with the gh attestation verify --predicate-type https://spdx.dev/Document/v2.3 command the workflow itself uses, and a note on the human-readable drydock-${TAG}.image.spdx.json copy attached to each release. No workflow change — the SBOM and build-provenance attestations, and their in-workflow verification, already met the house SLSA Build L2 / SBOM-attestation standard; only the user-facing documentation was missing. (#759)
Dropped the trivy qlty plugin in favor of Grype. Grype already owns the vuln-scanning surface (.github/workflows/security-grype.yml); trivy's qlty-plugin config driver only ever ran its DS*/KSV* Dockerfile misconfiguration checks, which checkov already covers (the checkov:skip=CKV_DOCKER_3 comment on Dockerfile line 1, with matching rationale in .trivyignore.yaml for the standalone trivy binary the image ships at runtime — a separate concern, untouched here). Removed the [[plugin]] name = "trivy" block and the trivy:DS002/trivy:DS-0002 triage entry from .qlty/qlty.toml; trufflehog stays as-is. (#753)
Faster initial load on the login screen. The app's logo asset had drifted 4x oversized and was still being base64-inlined into the parse-blocking entry chunk (~640KB, reachable before authentication). It's now correctly sized and served as a separately cacheable file, cutting the entry chunk's gzipped size roughly in half and no longer re-downloading on every release. (#805)
BREAKING: the Home Assistant MQTT topic layout now includes an agent/<name> segment by default for agent-owned containers. Announced in v1.5.x and scheduled for v1.7.0. Multi-agent deployments previously collided on shared watcher names; the corrected layout scopes topics, sensor counts, and discovery cleanup per agent. Home Assistant references to the old topic paths must be re-pointed. Set DD_NOTIFICATION_MQTT_<name>_HASS_AGENTTOPICSEGMENT=false to keep the old layout temporarily. Drydock does not retroactively clean up the old, pre-upgrade discovery entities. They remain retained on the broker and show up in Home Assistant as orphaned, frozen-at-last-state duplicates until manually removed.

Fixed

Dependency-group admission no longer loses restart context when an upstream member is rejected. A restart-only dependent whose upstream was rejected during admission was previously misclassified as a normal update in wave 0 instead of being restarted after its dependency finished (or skipped if it never ran). The original admission batch is now carried through dispatch and API annotation so rejected members no longer corrupt the graph for their accepted dependents. (#718)
Docker Compose updates no longer carry forward stale image-inherited environment defaults. A compose refresh could leave an old image's default environment value in place after updating to a new image with a different default, even though explicit runtime overrides on the container were always respected correctly. (#736, closes #734)
Stashed update-policy overrides now survive drydock's own self-update. The updatePolicyRetentionCache that lets a recreated container inherit its predecessor's policy overrides lived only in process memory and was silently lost on the exact restart caused by a self-update. It now persists write-through to the store and rehydrates at startup, mirroring the same fix already applied to the maturity-clock cache. (#743, fixes #565)

Fixed

Drydock no longer reports "Up to date" when an update check failed. A registry error while verifying a candidate's digest discarded the newer tag that had already been found, and the UI presented the result as a confirmed "no update available." A check that could not complete now surfaces as an explicitly unknown status instead of a negative answer. This also covers transient registry failures reported separately. (#814, #808)
A single malformed container no longer zeroes out an entire agent inventory sync. AgentClient.processAuthoritativeContainers() — the loop driving both the standard-mode handshake (_doHandshake(), every initial load and reconnect) and edge-mode handleContainerSync() (every dd:container_sync frame) — had no per-container error isolation, unlike the sibling SSE watcher-snapshot path. One container that threw while building its report (for example, during controller-side SBOM document offload) aborted the whole batch before any container reached the store, so a single bad container could make an otherwise-healthy fleet's inventory sync silently report zero containers. Each container in the batch is now processed independently, matching the existing isolation pattern in handleWatcherSnapshotEvent(): a failure is logged with the container id and the batch continues with the rest. A batch where every container fails now also logs a single proportionate warning, since a silently empty result was what let this go unnoticed for a release cycle. Found while investigating #802; this closes the isolation gap but does not confirm the SBOM path as that issue's trigger — Portwing does not currently send security.sbom on its container wire format, so #802 remains open pending its actual root cause.
Nested OCI image indexes now resolve to the real image manifest. An image whose per-platform entry is itself an index, which is what Buildx produces when SBOM or provenance attestations are enabled, failed with Unexpected error; no manifest found and left the container unable to complete a digest check. Drydock now follows the nested index to the platform's actual manifest, bounded to three levels, and correctly ignores the attestation manifest alongside it. (#814)

Removed

Retired the self-hosted star-history provider and the Warpchart embed that replaced it, in favor of a committed SVG chart. The self-hosted /api/star-history route read process.env.GITHUB_TOKEN at request time, and since that variable was always unset in production, it served a placeholder "check back shortly" SVG at HTTP 200 instead of erroring. Nothing ever went red, so the chart silently never rendered for as long as the route existed. Deleted the route handler, its SVG-rendering and request-resolution libraries, and the homepage StarHistory card. Warpchart was adopted next as its replacement, then retired again days later once that decision reversed. Both now give way to docs/assets/star-history.svg, a chart built from GitHub's own stargazer timestamps and committed straight into the repo: it needs no runtime secret and cannot return a placeholder at HTTP 200, so a stale chart is visible and a missing one is a broken image, not a silent failure. Regeneration is handled by a scheduled shared workflow landing separately. All seven README language variants now point at the committed SVG instead of an external chart service.
BREAKING: trigger-excluded and trigger-not-included are now hard blockers, ending the v1.5.0–v1.6.x soft-severity grace period (spec-6.0.1-action-policy.md, slice 6 — see deprecations). Previously, a container filtered out by dd.action.exclude/dd.action.include (or the legacy dd.trigger.* labels) showed a Trigger filtered/Trigger excluded pill but still let a manual Update click through after a confirm-modal warning. As of this release both reasons are hard: the Update button is locked, and the API rejects a manual update request with the blocker's 409 message instead of queuing it — matching every other hard eligibility gate. This ships together with, and not before, the per-action execution policy (dd.action.auto label, AUTO=onauto trigger mode) landed across slices 1-5, so the manual-only escape hatch isn't removed without a replacement path: an operator who wants a container to stay manually updatable while excluded from automatic dispatch now expresses that with AUTO=onauto plus dd.action.include (or dd.action.auto), rather than relying on the soft bypass. AUTO=oninclude's existing meaning is unchanged — a matching dd.action.include label still grants both manual and automatic access under that mode.
Removed the unversioned GET /api/auth/methods alias. Deprecated since v1.6.0 and scheduled for removal in v1.7.0, it now returns 410 like the rest of the unversioned /api/* surface. Use GET /api/v1/auth/status instead.

Removed

Removed curl from the Docker image. Deprecated since v1.5.x with removal scheduled for v1.7.0. The image's own healthcheck has used the compiled /bin/healthcheck binary since v1.5.0; user-defined HEALTHCHECK overrides that shell out to curl will stop working and should switch to test: /bin/healthcheck $${DD_SERVER_PORT:-3000} (the doubled $ is compose escaping) or drop the override. Drydock now also logs a startup warning naming the container when it detects that its own healthcheck override still shells out to curl (in addition to the existing UI banner).

Security

Scoped the Grype image gate around CVE-2026-14456 (openssl/libssl3/libcrypto3, Alpine 3.24) pending an upstream fix. ci-verify.yml's grype-image job started failing --fail-on high after this CVE (CWE-770, unbounded memory growth in OpenSSL's QUIC server incoming-channel-queue handling) entered the Grype DB; the packages are pinned at 3.5.7-r0 and no Alpine branch — 3.24-stable or edge — has repackaged the upstream 3.5.8 fix yet. Drydock links these libs only as a TLS client (the openssl CLI and Node's own TLS stack) for registry queries and the healthcheck, never as a QUIC/HTTP3 server, so the vulnerable path isn't reachable; OpenSSL's own advisory rates the flaw Low, versus the generic CVSS 7.5 the gate scores. Added a CVE-scoped (not location-scoped) .grype.yaml ignore entry for exactly this CVE on these three package names, so a different HIGH/CRITICAL finding in openssl still fails the gate. Two removal triggers: bump the Dockerfile's pinned openssl= version once Alpine ships >= 3.5.8 for 3.24, or 2026-11-17 (90 days out) regardless.
Closed a remote-property-injection path in the container list's URL query sync. ContainersView's query-normalization helper wrote URL-controlled keys onto a plain object with bracket notation, which could in principle let a crafted __proto__ query key reach the object's prototype chain. It now accumulates entries in a Map and materializes the result with Object.fromEntries, closing the vulnerability class rather than just suppressing the scanner finding. (#750)

Documentation

Retroactively documented the WUD_AGENT_SECRET/WUD_AGENT_SECRET_FILE removal. v1.6.0-rc.1 silently dropped agent mode's DD_AGENT_SECRET ?? WUD_AGENT_SECRET fallback (and the _FILE equivalent) with no deprecation period, and it was never recorded in the v1.6.0 changelog entry or deprecations at the time. An agent configured with only the legacy WUD_AGENT_SECRET went from authenticating to a startup failure whose message doesn't name the variable actually set. See the new deprecations entry.

1.7.0-rc.1 — 2026-08-14

Added

Edge Portwing polling cadence is configurable (#688). DD_PORTWING_POLL_INTERVAL sets the controller-owned Edge container-refresh interval in positive integer seconds; the authenticated welcome frame and reported agent metadata use the same value, while absent or invalid values retain the 300-second default.
Installable PWA support (Roadmap Phase 6.9). Drydock is now an installable Progressive Web App via vite-plugin-pwa: a web app manifest (Drydock, standalone display, theme/background color matched to the One Dark default --dd-bg, 192/512 icons plus dedicated maskable variants with safe-zone padding) and an auto-updating service worker (registerType: 'autoUpdate') that precaches the SPA shell so the dashboard boots offline. /api/** is explicitly excluded from all service-worker handling — no navigation fallback, no runtime caching — so a live dashboard never serves stale API data from cache; those requests always hit the network and surface a normal error if it's unreachable. A dismissible install banner (new InstallBanner component, following the existing AnnouncementBanner pattern) listens for the browser's beforeinstallprompt event and offers a one-click install, with the dismissal persisted under a versioned localStorage key. iOS home-screen install is supported via apple-mobile-web-app-capable and the existing apple-touch-icon. The backend's static UI server now serves sw.js with Cache-Control: no-cache so a new deploy is never masked by a browser-cached service worker script.
Clickable port links in the container list and detail views. Each host-published port in a container's details.ports now renders as a link (opened in a new tab, rel="noopener noreferrer") instead of inert text — in the side panel, the full-page detail tabs, and new opt-in "Ports" columns/rows in the table and card views. The scheme is auto-detected from the container-side port (443/8443 → https://, everything else → http://); the link target host prefers the port's own bound HostIp when it's a real address (not 0.0.0.0/::/::0), falling back to the agent's configured host for agent-watched containers, or the browser's own hostname otherwise. Internal-only (unpublished) ports still render as plain text. A new dd.port.label container label lets you attach a friendly name to a specific port (dd.port.label=80=Web UI,443=Admin Console) shown in place of the raw hostPort->containerPort/protocol mapping.
Container uptime, with a live-refreshing display. The existing details.startedAt field (from Docker's State.StartedAt) now also drives an opt-in "Uptime" tooltip showing the exact start timestamp in the container list, and a live "Up …" indicator in the card view's footer — both refresh on a timer and update immediately on SSE container-state changes, matching the full-page detail view's existing uptime display.
Keyboard shortcuts. / focuses the search bar from anywhere (unless focus is already in a text input), Escape closes the search bar, and ? opens a new shortcut-reference overlay listing the available shortcuts. A / hint now sits next to the existing ⌘K hint on the sidebar search button.
Container dependency ordering — data model and detection (v1.7, discussion #219). New dd.depends_on (comma-separated container names) and dd.depends_on.action (update or restart, default update) container labels. When dd.depends_on is absent, drydock detects dependencies automatically from a compose-managed container's own depends_on key (both short-form arrays and long-form objects; the condition: key is not yet consulted). A present label always overrides compose detection entirely rather than merging with it. Self-references and unknown compose-service targets are dropped with a logged warning, never a hard error; both fields are re-derived from the container's labels/compose file every watch cycle rather than persisted independently, so they self-heal automatically across container recreation. This lands the data model and detection only — using the resulting graph to order updates/restarts is a separate, later change.
Container dependency ordering — pure graph engine (v1.7, discussion #219). New app/dependencies/dependency-graph.ts: buildDependencyGraph resolves each container's detected dependsOn names against the rest of the fleet (dropping unknown targets and cross-agent edges with a logged warning, never a hard error — cross-host dependency chains remain out of scope for v1.7), and topologicalSort runs Kahn's algorithm to produce deterministic topological "waves" — arrays of containers safe to dispatch in parallel, tie-broken alphabetically for stable output. A dependency cycle is never a deadlock: cycle members are grouped and scheduled together as one unordered wave, while any non-cycle container downstream of that cycle still resolves correctly in its own later wave. Pure, dependency-free, and unwired — no watcher, trigger, or dispatch behavior changes yet; this only lands the engine that a later change will use to order updates.
Container dependency ordering — execution integration (v1.7, discussion #219). Accepted bulk container updates now dispatch wave-by-wave through runAcceptedContainerUpdates (app/updates/request-update.ts) instead of all at once, so a container never starts updating before every container it depends on has finished. A dependsOnAction: 'restart' dependent is admitted through the same admission gates as any manual request (widened updateAvailable check) and, once its dependency finishes updating, is restarted rather than re-pulled via the new restartDependentContainer primitive (app/updates/dependency-restart.ts); operations that never got a chance to run because an earlier wave failed land in a new skipped-dependency status/phase instead of silently vanishing. The Docker Compose trigger (app/triggers/providers/dockercompose/Dockercompose.ts) reorders multi-service docker compose up invocations by dependency order (sortMappingsByDependencyOrder) so compose itself never fights the same ordering. Maintenance-window batches (app/triggers/providers/Trigger.ts's runAcceptedUpdateBatch) cascade dependents through the same wave logic once their window opens.
Container dependency ordering — API exposure (v1.7, discussion #219). The container list response gains per-container dependencyCount/dependentCount badge counts. New endpoints: GET /api/v1/containers/dependencies returns the full resolved dependency graph (nodes, edges, detected cycles, unresolved targets, cross-host-ignored edges); POST /api/v1/containers/:id/update-chain-preview dry-runs the topological waves for the dependency chain rooted at a container without dispatching anything; POST /api/v1/dependency-groups/:rootId/update bulk-accepts every container in that chain, annotated with the wave index it will actually run in. The preview and dispatch endpoints call the exact same buildDependencyGraph/topologicalSort pair over the same input set, so the preview can never drift from what an accepted update actually runs.
Container dependency ordering — UI (v1.7, discussion #219). The container list now carries dependencyCount/dependentCount through to the UI Container type. A new "Update dependency chain" action (confirmDependencyGroupUpdate in useContainerActions) previews the resolved update waves for a container's dependency chain and shows them in a confirm dialog — including any detected cycle or unresolved-target warnings — before bulk-accepting the chain through the new dependency-groups endpoint. A dedicated dependency-hierarchy grouping view for the container list is deferred to a follow-up.
Debounced container discovery (#156). Docker briefly exposes transient rename aliases while a container is being recreated; drydock previously registered whatever it saw the instant listContainers returned it, so a container could momentarily register under its <hex-prefix>_<name> alias. First-seen containers (identified by Docker container ID, not present in the store) now enter a configurable "pending" state and must remain visible for a settling window — DD_WATCHER_{watcher_name}_DISCOVERY_SETTLE_MS, default 30000 (30s), 0 disables settling — before they're added to the store, triggers, or the API/UI; pending containers are visible in debug logs only. A deduplicated follow-up watch is scheduled for the earliest pending deadline, so an event-discovered container still registers on time when no further Docker event arrives before the next cron scan. If a pending container is renamed mid-window it registers under the final name once settled; if it disappears mid-window it's silently discarded (debug log only). Containers already known to the store are unaffected and continue to update immediately — settling applies exclusively to first-seen containers, so a same-ID recreation is never blocked from updating for 30 seconds. This complements, and does not replace, the unconditional hex-prefix alias stripping shipped in v1.5 for the same issue (getContainerName/canonicalizeContainerName in app/watchers/providers/docker/docker-helpers.ts, and the name-shape-triggered transient-alias suppression in filterRecreatedContainerAliases) — that mechanism is keyed off the container's name looking like a recreate alias, while the new settling window (filterPendingDiscoveries in app/watchers/providers/docker/container-init.ts) gates any first-seen container regardless of name shape.

Changed

Dependency-graph engine: iterative cycle detection, single-pass wave resolution, and indexed candidate lookup (v1.7, discussion #219). stronglyConnectedComponents (app/dependencies/dependency-graph.ts) no longer recurses — an explicit work-stack replaces the recursive strongConnect — so a single dependsOn chain or cycle around 5-10k+ containers no longer risks a stack-overflow RangeError; a fleet's dependency graph has no size bound a recursive implementation could safely assume. topologicalSort's cycle-resolution loop no longer rebuilds the remaining subgraph and recomputes strongly-connected components from scratch every round (O(N²) on fleets with several chained/dependent cycles) — a single upfront SCC computation plus one forward pass over the resulting condensation now produces byte-identical wave/cycle output in O(V+E). buildDependencyGraph's label and compose candidate lookups (findLabelCandidates/findComposeCandidates) are now backed by a once-per-call watcher/name and compose-project/service index instead of a full container-list scan per dependsOn entry.

Removed

BREAKING: The legacy DD_TRIGGER_* environment variable prefix and dd.trigger.include / dd.trigger.exclude container labels are removed (deprecated v1.5.0, warned at error level throughout v1.6.0, removed per the published v1.7.0 schedule in deprecations). Any DD_TRIGGER_* environment variable now fails startup outright — the error lists every detected variable next to its exact DD_ACTION_* (docker/dockercompose/command) or DD_NOTIFICATION_* (every other provider) replacement, plus the config migrate --source trigger command and a link to the deprecations page, so a config can be fixed in one pass. dd.trigger.include / dd.trigger.exclude container labels no longer resolve to anything — only the scoped dd.action.* / dd.notification.* labels are read; a container still carrying either legacy label logs a one-time error-level warning and keeps incrementing the dd_legacy_input_total{source="label"} counter (surfaced in the existing deprecation banner) so unmigrated fleets stay visible even though the label is no longer honored. The usesLegacyPrefix trigger-metadata field, the now-always-empty legacy-prefix tracking in app/configuration/index.ts, and the superseded startup warning are removed along with it. The drydock config migrate --source trigger CLI is unaffected — it's a standalone, offline config-file rewriter and remains the recommended migration path.

Security

The 2026-08-13 security pass closes six resource and credential-exposure gaps. Login admission now caps concurrent password verification before Argon2 runs; standard agent JSON requests have time, body, response, and redirect bounds; unterminated agent SSE events have a finite buffer; container log downloads and initial WebSocket history have finite line/byte limits; slow local log viewers are disconnected; registry data requests refuse redirects; and command/hook strings are redacted from component APIs and execution logs. The dated findings, evidence, and validation record are in security_best_practices_report.md.
Added a root .trivyignore.yaml suppressing AVD-DS-0002 (Dockerfile missing USER) with the same rationale already documented for the Dockerfile's checkov:skip=CKV_DOCKER_3 comment and the existing qlty trivy:DS-0002 triage rule: the entrypoint drops privileges at runtime via su-exec (Docker.entrypoint.sh), so no static USER instruction is needed.
Service-worker NetworkOnly rule for /api/** now actually matches. ui/vite.config.ts's runtimeCaching entry used a ^-anchored pathname regex (/^\/api\//), but workbox-routing tests a RegExpRoute's urlPattern against the full url.href (always starting http:///https://), never the pathname alone, so the rule could never match and silently fell through. It was harmless today only because no other runtimeCaching rule exists to catch the fallthrough — any future catch-all caching rule would have started silently caching authenticated /api responses. Replaced with an exported isApiRequest match-callback function that tests url.pathname.startsWith('/api/'), so the rule actually engages.
Dependency-group bulk update now requires destructive-action confirmation and binds to its own preview (v1.7, discussion #219). POST /api/v1/dependency-groups/:rootId/update could update or restart every container in a resolved dependency chain with no confirmation step and no binding to whatever chain the UI last previewed — a container added to the chain between preview and confirm was silently swept into the update. The route now requires the X-DD-Confirm-Action: dependency-group-update header, matching the existing container-delete pattern, and accepts an optional expectedContainerIds array in the request body; when present, a live chain that no longer matches it exactly (order-insensitive) is rejected with 409 and the actual current chain, instead of running against a chain the caller never saw.

Fixed

Edge exec completion reasons now reach the controller-side consumer (#635). String exec_end.reason values are forwarded through the internal startExec end callback; sessions are removed before consumer code runs, and a throwing callback cannot leak state or stop disconnect cleanup.
Controller-owned Portwing containers now use the controller's configured registry identity before native checks (#687). Complete Docker-transport inventory and event records are normalized to the canonical registry name, URL, credentials, and image identity, so watch-now no longer delegates registry work to Portwing's intentional 501 stub; traditional agents and partial events keep their existing behavior.
Manifest created metadata failures no longer become successful missing dates (#606). Exhausted 4xx, 5xx, network, and non-object failures now propagate the original failure; only 301/302/303/307/308 responses on optional manifest/blob metadata may omit created, preserving an already resolved digest while redirect following stays disabled.
Demo site favicon now matches the refreshed branding. The v1.5.1 brand refresh (#439) moved the website to the cropped whale "headshot" icon and the app UI followed, but demo.getdrydock.com kept showing the old full-body whale: its stale favicon.svg — which modern browsers preferred over the PNGs — was never replaced. The demo now ships the same headshot icon set as the website and app UI, the favicon.svg is removed, and the icon links carry a ?v=2 cache-buster so browsers re-fetch instead of serving the aggressively cached old icon. (#689, forward-ported in #690)
Audit nav icon no longer renders blank for the Lucide icon-library preference, and the icon bundle no longer silently drops aliased or renamed icons at image build time. PR #668's icon-bundle regeneration against tabler 1.2.38 dropped lucide:history from ui/src/boot/icon-bundle.json because the installed @iconify-json/lucide (1.2.121) demoted history to an alias of rotate-ccw-clock; ui/scripts/extract-icons.mjs's bundler only ever looked up plain icon entries, never aliases, so every image built since that pin shipped without it (no network fallback, since the iconify API module is offline-only). The extractor now resolves alias chains (parent-following, depth-capped at 5), refusing any alias that carries a rotate/hFlip/vFlip transform the body-only bundle can't represent. ui/src/icons.ts's iconMap.audit.lucide is repointed to lucide:rotate-ccw-clock directly (the same clock-with-counter-clockwise-arrow concept fa6-solid:clock-rotate-left/ph:clock-counter-clockwise already use for the same entry) rather than relying on alias resolution for that one. A bundle-consistency sweep against the currently installed iconsets turned up more references that never resolved against the locked collections at all: iconoir:history → iconoir:clock-rotate-right, iconoir:gitlab → iconoir:gitlab-full, iconoir:stack → iconoir:multiple-pages, lucide:more-vertical → lucide:ellipsis-vertical, iconoir:key-alt → iconoir:key, plus four fa6-brands:* icons that were never bundled because @iconify-json/fa6-brands (now exact-pinned as a devDependency) was missing from ui/package.json entirely — all fixed or backfilled so every <library>:<icon> reference in iconMap now resolves. Two guard tests now cover this class of regression: ui/tests/icons.spec.ts asserts every iconMap entry has a matching icon-bundle.json key, and ui/tests/boot/icon-bundle.spec.ts independently asserts every 'prefix:name' reference found in icons.ts exists in the generated bundle with a body.
Compose-derived dependency detection now works when drydock itself runs in a container (v1.7, discussion #219). resolveComposeDependsOn (app/dependencies/compose-dependency-resolver.ts) read a container's compose file at its host-side label path — which doesn't exist inside drydock's own container filesystem unless a bind mount happens to line up 1:1 — so most non-trivial layouts silently detected zero dependencies. Host compose-file paths are now translated through drydock's own bind mounts (reusing the existing Docker Compose trigger's translation logic, ComposePathBindMounts.ts, cached per Docker API instance) before being read; when none of a container's configured compose file paths can be translated and read, that now surfaces as a single warning naming every path tried instead of a silent empty result.
Dependency-group update confirm dialog now binds to what it actually runs, and marks restart-only members (v1.7, discussion #219). Pairs with the destructive-confirmation entry above: the confirm dialog previously fetched a wave preview but never bound the eventual dispatch to it. confirmDependencyGroupUpdateState now forwards every previewed container id as expectedContainerIds on accept and surfaces a distinct "chain has changed" toast on a 409 divergence response instead of a generic failure message or a silent retry. The wave list in the confirm message now suffixes each restart-kind member with (restart) so restart-only dependents are visually distinguished from update targets before confirming.
Update age and hot/mature/established classification now always use the same trust-aware clock as the maturity gate itself (#556). The gate (resolveMaturityClock in app/model/maturity-policy.ts) already checked result.publishedAtTrusted before trusting a registry's publishedAt, falling back to updateDetectedAt/firstSeenAt otherwise — but three other call sites computed age independently and skipped that check: getRawUpdateAge (app/model/container.ts, feeding container.updateAge/updateMaturityLevel) blindly Math.min'd firstSeenAt and result.publishedAt; app/api/container/update-age.ts's uncached fallback ran its own three-way blend; and the UI's container-mapper.ts/useContainerPolicy.ts fallback branches (used whenever the eligibility payload has no active maturity-not-reached blocker to read the resolved clock off of) and the age tooltip formatter hand-rolled an updateDetectedAt-only heuristic. All four now delegate to the shared resolver (getUpdateAgeMs on the app side, a ported resolveMaturityClock mirror on the UI side) instead of re-deriving it. Behavior change: an untrusted early publishedAt (e.g. Docker Hub/GHCR OCI build dates on other registries, or any pre-#-trust-flag data) is no longer blended into the displayed age — containers whose updateAge/maturity badge previously looked artificially older can now show a smaller age and flip from mature/established back to hot; ?sort=age and ?maturity=hot|mature|established bucketing/ordering shift accordingly. This is the intended, correct direction (fail-closed: an untrusted date is never trusted for display any more than it is for gating) but is user-visible. No documented OpenAPI field changed — updateAge/updateMaturityLevel/updateDetectedAt/firstSeenAt/result.publishedAt* were never part of ContainerResource's documented schema (additionalProperties: true).
GHCR version-history pagination no longer silently caps out at 1,000 versions (#556). fetchVersionsPagedForOwner used to guess "another page exists" from versions.length === perPage and gave up after a hardcoded 10 pages, so any GHCR package with more than 1,000 published versions (routine for a CI-heavy repo doing per-commit/nightly tags over a couple of years) silently lost trusted publishedAt lookups for older tags with no operator-visible signal. Pagination now follows the literal Link: rel="next" URL GitHub's REST API returns (RFC 5988) — correct on the exact boundary the length heuristic got wrong — against a much higher, configurable ceiling (DD_GHCR_VERSIONS_MAX_PAGES, default 500 pages / 50,000 versions). Hitting that ceiling while a next page still exists now logs a warn distinguishing "truncated" from "confirmed absent"; the string | undefined return contract is unchanged, so a truncated scan still fails closed (no trusted publishedAt), it just stops being invisible when it happens.
updateLifecycleCache now survives drydock's own self-update instead of being wiped by it (#556). The maturity-clock carry-forward cache (app/store/container.ts) that lets a recreated container inherit its predecessor's updateDetectedAt/firstSeenAt/maturityGatePendingSince lived only in a bare process-memory Map, invisible to the SIGTERM shutdown() handler's store.save() flush — every collection except this one got persisted. Since a drydock self-update is definitionally a cross-process container recreation (recreate action → SIGTERM → new process), the stash was reliably lost on the exact restart that needed it, silently re-stamping updateDetectedAt as "now" and restarting any in-progress maturity soak. A new app/store/update-lifecycle-cache.ts module (modeled on the existing name-bindings.ts persistence precedent, which solved the identical bug class for the agent name→key binding cache) mirrors the in-memory cache into a LokiJS collection: write-through on stash, delete-through on consume/expiry/signature-mismatch and on size-based eviction, and a new rehydrateUpdateLifecycleCacheFromStore() repopulates the Map from non-expired persisted records once at startup, right after the collection is created. No new flush hook was needed — LokiJS collection writes are synchronous, so the existing store.save() call already picks up the persisted cache along with every other collection.

1.6.0 — 2026-08-11

Consolidates the 1.6.0-rc.1 … 1.6.0-rc.13 prereleases. Users upgrading from 1.5.2 get everything below; users already on 1.6.0-rc.13 receive no additional runtime changes.

Added

Portwing edge and agent integration matures into a first-class transport. Portwing 0.9.0+ agents work over either inbound Standard HTTP or dial-out Edge WebSocket transport, and drydock 1.6.0-rc.11+ recognizes Portwing's own Docker watcher marker to run native registry checks plus single/batch Docker updates controller-side (#632, #637, Portwing #76). Standard-mode agents can sign every request with Ed25519 (DD_AGENT_{name}_AUTHMODE=ed25519, upgraded mid-series to signature version 2) instead of sending a shared secret. Edge agents stream live container logs and process deletes over the existing WS tunnel with correlated per-request IDs, pick their own display name (bound to their signing key so it can't be squatted), and are proactively disconnected when they stop answering pings. The portwing/1.0 edge endpoint is enabled by default (DD_EXPERIMENTAL_PORTWING=false remains an emergency disable), backed by a real fleet-soak workflow that runs signed Portwing processes against production infrastructure on every PR and on a weekly schedule.
Declarative, three-tier update policy with a maturity stabilization gate (Discussion #307, Discussion #406, #320). Containers can declare dd.updatePolicy.* labels (maturity mode/min-age, skip tags, skip digests) with watcher-level defaults and UI/API overrides, resolving through label → watcher default → persistent override precedence and surviving agent refresh, container recreation, and error-state rebuilds. A candidate held back by maturityMode: mature is visible immediately with a live minute-by-minute countdown to its unlock time, and a new maturity-cleared notification (backed by a 5-minute background sweep) announces the moment it clears instead of waiting for the next scan. Update Now remains available as an explicit soft-policy override.
Per-rule notification templates, bell preferences, and a new health-status event (Discussion #205, Discussion #198, Discussion #210). The Notifications view can override title/body/batch-title text per rule and provider with a live preview, and independently control which event categories reach the in-app bell — including a new container-unhealthy rule (disabled by default) that fires off Docker health-check transitions. Bidirectional MQTT lets Home Assistant's Install button trigger a real drydock update through the normal eligibility/dispatch path, rate-limited per container and audited. dd.action.*/dd.notification.* labels are now strictly scoped to their own trigger category instead of one silently gating the other (#494).
Global update mode and an actionable Update Status panel (Discussion #325). Settings → General now selects one server-wide mode — notify, manual, or auto — and the container side panel and detail view replace the old eligibility-badge stack with plain-language status plus a structured, deep-linking condition list covering all 16 eligibility reasons. Existing installs migrate to auto on upgrade so configured automatic updates keep working; fresh installs default to manual (see Upgrade Notes).
Cross-device preference sync and a zero-dependency dashboard grid (Discussion #220, #281). An opt-in Sync across devices toggle stores the full UI preference set server-side per user and propagates changes to a user's other signed-in sessions in real time over SSE. The dashboard grid drops its grid-layout-plus dependency for a deterministic CSS Grid implementation with edit-mode drag/reorder, bounded resizing, touch support, and per-breakpoint layout persistence. Audit, Security, Servers, and Watchers gain the same per-view column show/hide picker Containers and Agents already had.
More signal, less noise in update visibility. A pinned tag now surfaces the best newer same-family tag as a purely informational badge without making it an actionable update candidate (#498); a new Version Update filter isolates real semver bumps from digest-only churn on fleets that rebuild images daily (#538); and a startup warning flags minute-precise maintenance-window crons that only open their window for one minute per matching hour instead of the intended full range (Discussion #639).
Opt-in wud-card/Homepage compatibility endpoints. With the unversioned /api/* alias removed (see Removed, below), DD_COMPAT_WUDCARD=true (default false) mounts a narrow compatibility layer covering exactly the four endpoints the Home Assistant wud-card integration and Homepage's whatsupdocker widget call, reshaped into the bare-array response those integrations expect. (Discussion #469)

Changed

Every major list view is responsive and consistent. Containers, Agents, Notifications, Security, Triggers, Watchers, Servers, Registries, Audit, and Auth now share one DataTable with a persisted per-view table⇄card toggle that auto-reflows to cards below ~640px; card views gained a sort control; and Source/Release-notes/Registry resource shortcuts render through one consistent 44px toolbar everywhere they appear (#498, Discussion #295). The Dashboard and Containers views no longer overflow horizontally on narrow phone screens. Update-status vocabulary was overhauled ("Digest update", "Security hold", neutral Major/Minor/Patch badges for informational-only updates), pinned reverted to being a tag property rather than an update status, and the maturity panel now shows exactly one countdown clock instead of two that could disagree with the gate (#556). Table polish across the series fixed pinned-column overlay clipping, a registry error overwriting the visible tag, sticky-column/auto-hide width miscalculation, and a clipped icon column.
Trigger taxonomy migration reaches its final warning stage. Every DD_TRIGGER_* variable and the deprecated dd.trigger.include/dd.trigger.exclude labels still work in v1.6 but now log at error level ahead of removal in v1.7; use DD_ACTION_*/DD_NOTIFICATION_* and category-scoped labels, or run config migrate --source trigger.
Less duplicate work on the hot paths. Registry tag-list requests are shared across containers within a poll instead of being repeated per container; scheduled security scans and the shared log viewer avoid redundant store writes and re-renders; and transient registry network errors (timeouts, connection resets, DNS blips) now retry with backoff before a watch error is recorded.
Base image bumped from Alpine 3.21 to 3.24, later rebased onto node:24-alpine at Node 24.19.0 with a matching Trivy build-stage bump to 0.73.0, closing several HIGH/MEDIUM CVEs in the underlying image (#682).

Deprecated

GET /api/auth/methods and the legacy GET /auth/strategies response shape are deprecated in v1.6.0 (removal in v1.7.0 and v1.8.0 respectively). Both now log on every request and return Deprecation/Sunset headers pointing callers at canonical GET /api/v1/auth/status. See DEPRECATIONS.md.

Removed

Unversioned /api/* and WS /api/log/stream aliases are gone, returning 410 Gone (or a rejected upgrade) instead of serving the request. This is a breaking change for integrations that hardcode the unversioned base path, notably the Home Assistant wud-card integration and Homepage's whatsupdocker widget — use the DD_COMPAT_WUDCARD shim above, or migrate to /api/v1/* directly.
Legacy v1.4-era authentication compatibility. Basic auth now accepts only argon2id hashes ({SHA}, APR1/MD5, crypt, and plain-text hashes fail validation), and OIDC discovery now requires https:// with no insecure HTTP workaround.
Legacy WUD configuration aliases, obsolete watcher switches, and legacy trigger-template variables. WUD_* environment variables and wud.* labels are ignored (the migration CLI still recognizes them so existing files can be rewritten); DD_WATCHER_<name>_WATCHDIGEST/WATCHATSTART are no longer configuration keys (use dd.watch.digest=true; startup watches always run); and $id/$name/$watcher/$kind/$semver/$local/$remote/$link/$count no longer populate in trigger templates.
Kafka clientId and token-only public-registry compatibility handling. Kafka validation accepts only lowercase clientid, and malformed public-registry configurations (a bare PUBLIC_TOKEN with no PUBLIC_LOGIN) now fail closed instead of silently falling back to anonymous pulls.

Fixed

The maturity soak clock, update-policy overrides, and digest identity now survive the paths that used to reset them. Container recreation, agent report ingestion, watch errors, concurrent security scans, and manual rechecks each previously had at least one path that could silently restart a maturity countdown, drop a stored policy override, or roll back freshly detected update state — every one of those is closed, including the store's digest anchor now self-healing from a stale, multi-entry RepoDigests ordering instead of trusting whichever entry happened to land first (#669). Notification dedup no longer fires a duplicate one-shot notification when a manual recheck returns drifted metadata for the same candidate, and containers with no available update are no longer ranked or filtered as though they had one.
Agent- and edge-specific reliability fixes. The container list no longer flashes a false "Agent Mismatch" during an agent's brief re-registration window; start/stop/restart/rollback return an explicit 501 instead of an ambiguous 404 when an agent lacks lifecycle transport (#637); WebSocket log streams accept anonymous-auth sessions; and edge memoryGb now reports binary GiB instead of decimal GB.
Operational fixes. The Star History chart is now self-hosted after the third-party chart services' outage (#671); the icon bundle no longer silently drops icons that became iconify aliases mid-series; GET /api/v1/containers/backups is reachable again (it was shadowed by the container router's own root handler); infrastructure-mode self-updates no longer fail to spawn their helper container on Docker Hub installs (#644); unchanged pending updates stop bloating the audit log on a timer; and the app and demo favicons match the v1.5.1 brand refresh (#439).

Security

Anonymous access fails closed on upgrade, not just on fresh installs. An unconfigured or unconfirmed-anonymous instance now rejects protected API requests with 401 instead of showing an open dashboard; set DD_ANONYMOUS_AUTH_CONFIRM=true to keep an intentionally open instance working.
Transport and session hardening. HTTP notification triggers are hardened against SSRF (blocked cloud-metadata/link-local targets, contained redirects); WebSocket upgrades validate the complete origin instead of a host substring; the session cookie is renamed to drydock.sid, signing every existing user out once; /store and its contents are created owner-only (0700/0600); and icon CDN sources are pinned to exact upstream revisions instead of floating tags.
Ongoing dependency and base-image CVE sweeps ran throughout the series — notably Next.js 16.2.11 and React 19.2.8 on the documentation site, fast-uri/ip-address/brace-expansion override advances, and the Node 24.19.0 + Trivy 0.73.0 bump in the final release candidate. See the individual rc.* entries below for full CVE detail.

Upgrade Notes

Existing installs keep automatic updates; fresh installs start in manual mode. The new global updateMode setting defaults to manual only when drydock creates a settings record for the first time; an existing record with no updateMode migrates to auto, preserving pre-v1.6 behavior. Review Settings → General → Update mode after upgrading if you prefer notifier-only or manual-only operation.
A lone dd.action.include/dd.action.exclude (or dd.notification.include/dd.notification.exclude) label stops filtering the other trigger category. If you relied on the pre-v1.6 cross-category leak (#494), previously-suppressed notification (or action) triggers may fire once on the first scan after upgrading. drydock logs a one-time warning per affected container naming the missing label; set the matching label to the same value to restore the previous filtering.

Documentation

Podman/Docker socket security docs refreshed, and a full configuration/env-var and API/OpenAPI parity audit brought the docs and generated spec in line with the shipped /api/v1 surface, including the authentication component, registry webhook signature auth, notification outbox actions, bulk container updates, backup listing, and container summary/update-response details.

1.6.0-rc.13 — 2026-08-08

Security

Pinned js-yaml to 3.15.1 in the e2e workspace (override; transitive dependency) for GHSA-5p4m-2wfm-xmqj (CVE-2026-59870 backport gap: quadratic CPU consumption in !!omap resolution).
Pinned nanoid to 3.3.18 (override; transitive dependency of postcss in the root, app, apps/demo, apps/web, and ui workspaces, and of artillery in the e2e workspace) for GHSA-2v37-7h3g-55p8 (CVE-2026-67213) and, in the e2e workspace which was still on 3.3.12, also GHSA-28wg-ghj8-5hjv (CVE-2026-67214).
node:24-alpine base image bumped to Node 24.19.0, replacing the Node 24.18.0 image shipped in rc.12. Picks up Node's July 29 2026 security release, fixing 3 HIGH (CVE-2026-56846, CVE-2026-56848, CVE-2026-58043) + 5 MEDIUM CVEs that landed in 24.18.1 (#682).
Vendored aquasec/trivy build-stage pin bumped from 0.72.0 to 0.73.0, resolving 4 HIGH / 6 MEDIUM CVEs in its vendored Go dependencies: go-git (CVE-2026-71556), x/text (CVE-2026-56852), grpc (GHSA-hrxh-6v49-42gf), oras-go (CVE-2026-50151, CVE-2026-50163), and the Go stdlib (CVE-2026-39822) (#682).

Fixed

Icon bundle no longer silently drops referenced icons at image build time (#683). The bundle is regenerated from the locked @iconify-json packages during every Docker image build, but the extractor only looked up plain icon entries — lucide:history (the Audit navigation icon in the Lucide icon theme) became an alias in lucide 1.2.121 and vanished from shipped images, rendering blank. The extractor now resolves alias chains, and references that never existed in the locked collections are fixed: iconoir:history → iconoir:clock-rotate-right, iconoir:gitlab → iconoir:gitlab-full, iconoir:stack → iconoir:multiple-pages, and the Font Awesome brand glyphs (GitHub/GitLab/Google/Microsoft registry icons) gained the previously missing @iconify-json/fa6-brands package. A new test asserts every icon referenced in icons.ts exists in the committed bundle.
Star History chart is now self-hosted (#671). The homepage card and README embed rendered a broken image after api.star-history.com's global outage (their GitHub tokens rate-limited; starchart.cc also failing). A new /api/star-history route on the website fetches stargazer timestamps from the GitHub API server-side (optional GITHUB_TOKEN, edge-cached six hours with stale-while-revalidate, short-lived fallback SVG on fetch failure) and renders the chart in the site's own palette for both themes; the README uses a <picture> element with theme-matched variants. No third-party chart service remains in the path, and api.star-history.com is dropped from the site's CSP img-src.
Digest-update comparison no longer anchors on an arbitrary RepoDigests[0] entry (#669). A local Docker image can carry multiple repo@digest entries for one Image ID (pull/retag accumulation, no ordering guarantee); getRepoDigest blindly took index 0, so a stale or foreign-repo entry landing first anchored the whole digest-update pipeline to the wrong manifest and produced a persistent digest-update false positive that survived applying the update. getOrderedRepoDigests (app/watchers/providers/docker/docker-helpers.ts) now returns every RepoDigests entry whose repo component matches the container's own image reference, ordered, falling back to the full list only when nothing matches; the container model gained an optional image.digest.repoDigests field carrying that ordered list, re-derived from the live Docker image inspect on every discovery/refresh cycle. handleDigestWatch (app/watchers/providers/docker/image-comparison.ts) now walks that candidate list — a cheap raw-value check first, then a normalize-and-compare registry call per remaining candidate, skipping anchors whose manifest lookup fails — and re-anchors digest.repo to whichever candidate actually matched, so a store already poisoned with a stale digest.repo self-heals on its own. A genuine same-tag republish (no candidate matches) still flags an update exactly as before; if every candidate fails to normalize, the failure now propagates instead of being silently coerced into a false "no update".

1.6.0-rc.12 — 2026-08-04

Changed

Routine dependency maintenance across the root tooling, demo, UI, and website workspaces (#653, #654, #655, #656). The dependency-version guard's Next.js check became a 16.x floor instead of an exact pin so routine patch bumps stop tripping it (#664).
Crowdin translation sync (#665): one French container-component string corrected.

Fixed

"Agent Mismatch" no longer appears in the container list/SSE display during the brief window an agent's docker/dockercompose trigger is still (re)registering (#605). Eligibility is recomputed live on every read, and AgentClient._doHandshake() deregisters the agent's components before awaiting the /api/triggers fetch and re-register. A read in that window found zero triggers for the agent and computeUpdateEligibility raised a hard agent-mismatch blocker, disabling the Update button, even though nothing was actually misconfigured — the condition self-corrected once registration finished. agent-mismatch now downgrades to a soft blocker (button stays enabled) on display surfaces whenever the container's own agent is mid-registration, per the new AgentClient.isRegisteringComponents flag (true only for the deregister→re-register span, not the whole reconnect backoff). Update admission (app/updates/request-update.ts) is unaffected and stays hard/fail-closed throughout, so an update can never be enqueued through a wrong-agent trigger during that window.
WebSocket log streams no longer reject anonymous-auth sessions (#636). Both WS upgrade paths — the system log stream and the container log stream — gated on isAuthenticatedSession() requiring session.passport.user, which passport-anonymous never sets, so under DD_ANONYMOUS_AUTH_CONFIRM=true the log stream WebSocket always rejected the upgrade even though every REST endpoint worked. isAuthenticatedSession now also accepts the session when anonymous authentication is the registered mode.
Maturity clock: swallowed auth errors surfaced, per-container threshold respected (#604). getImagePublishedAt failures — including GHCR/LSCR 401/403 auth errors — now log at warn instead of debug, so the maturity gate's silent fallback from the registry publishedAt to updateDetectedAt is no longer invisible. getRawUpdateMaturityLevel (app/model/container.ts) and getContainerMaturityLevel (app/api/container/maturity-filter.ts) now resolve each container's own updatePolicy.maturityMinAgeDays before falling back to the global DD_UI_MATURITY_THRESHOLD_DAYS, matching the gate's own isUpdateSuppressed/isMaturityGatePending logic so the hot/mature badge can no longer disagree with the gate in the same API response.
Container start/stop/restart/rollback return an explicit 501 instead of an ambiguous 404 for agent containers without lifecycle transport (#637). POST /:id/start|stop|restart and POST /:id/rollback returned a bare 404 No docker trigger found for this container whenever the lookup missed, indistinguishable from "container not found" — for agent-owned containers this was the only signal the UI got. That lookup miss now returns 501 naming the likely cause (the agent's connection typically hasn't advertised usesControllerDockerTransport) when container.agent is set; non-agent containers still get the existing 404. This complements the native-transport support that shipped in rc.11 via #651, which closed #637's core gap — this is the remaining explicit-error half.

Security

brace-expansion, ip-address, and fast-uri overrides advanced to patched releases. brace-expansion moved to 5.0.9 in app/, ui/, and e2e/ (CVE-2026-69152, GHSA-rgw5-rvv9-x895); ip-address moved to 10.3.1 in app/ (CVE-2026-54272, CVE-2026-69192, CVE-2026-69198), pulled in transitively via express-rate-limit and mqtt → socks; fast-uri advanced from 4.1.1 to 4.1.2 in app/ and ui/ (host confusion via backslash authority introducer, CVE-2026-18446, GHSA-7p8r-x3mc-p8w7, superseding #658).

1.6.0-rc.11 — 2026-08-01

Added

Portwing controller-owned Docker watcher, update, and lifecycle transport (#632, #637, Portwing #76). Drydock 1.6.0-rc.11+ recognizes Portwing 0.9.0's exact Docker watcher marker (transport=docker-api, execution=controller, events=portwing) and runs its native registry checks plus single/batch Docker updates controller-side. Container start, stop, restart, update preview, and backup rollback actions use that same native Docker capability surface instead of failing because Portwing intentionally advertises no remote trigger. A loopback-only bearer-authenticated bridge carries Docker API calls through Portwing over Standard HTTP or Edge correlated request/response/stream messages, while Portwing remains the lifecycle-event source. Later raw Portwing inventory (updateAvailable=false, updateKind=unknown) preserves rather than erases the controller-enriched update state.

Changed

Standard Portwing Ed25519 requests now use signature version 2. The controller sends five authentication headers, including X-Portwing-Signature-Version: 2, and signs the exact origin-form request target (escaped path plus the unmodified raw query) instead of a decoded path without its query.

1.6.0-rc.10 — 2026-07-31

Added

Startup warning for minute-precise maintenance-window crons. DD_WATCHER_{name}_MAINTENANCE_WINDOW is matched minute-by-minute, so a fixed minute field like 0 2-6 * * * only opens the window for one minute per matching hour instead of the whole hour range — a common copy-paste trap reported in Discussion #639. The Docker watcher now logs a one-time warning at init when the configured window's minute field doesn't contain *, pointing at the fix (* 2-3 * * *). Step values like */5 are intentional and are not flagged.
"Learn more" link in the update confirm dialog for policy-blocked overrides. When overriding a soft-blocked update, the confirm dialog now links to the update-eligibility reasons reference, matching the link already shown in the Update Status panel. (Discussion #639)

Changed

Translations resynced from Crowdin (#620), refreshing the container-component and list-view catalogs across all 15 non-English locales, plus a fuller French pass over the agents, app-shell, common, and config catalogs.
Routine dependency maintenance across the app and UI workspaces (#614, #615, #617, #646): undici 8.9.0 (app) and the 7.29.0 UI override, express-rate-limit 8.6.1, @aws-sdk/client-ecr 3.1096.0, vue-i18n 11.4.8, knip 6.29.0, postcss 8.5.24, and the fast-uri security override advanced from 3.1.4 to 4.1.1 (both retain the CVE-2026-16221 fix; the dependency-version guard now rejects the still-vulnerable 4.0.0–4.1.0 range). Renovate no longer proposes @babel/core majors while Stryker's instrumenter requires Babel 7.

Fixed

Unchanged update-available audit entries are no longer re-recorded on a timer. Previously an unchanged pending update was re-written to the audit log every time the 1-hour dedupe window lapsed, bloating the audit log for fleets with long-pending updates (a fleet with 10 persistent pending updates wrote 140 rows in 24h). Audit rows are now recorded only on first detection or when the update target/kind changes. The RC-only DD_AUDIT_UPDATE_AVAILABLE_DEDUPE_MS variable is removed.
Maintenance-window documentation corrected and made findable (Discussion #639). Every documented example used a minute-precise cron (0 2-6 * * *) while describing it as an hourly range ("2am–6am daily"). MAINTENANCE_WINDOW is matched minute by minute, so that expression opens the window for one minute at the top of each listed hour and reports "closed" the rest of the day. Examples now use * 2-6 * * *, with a table of correct forms in Watchers. The update confirm dialog's own wording ("This update is currently policy-blocked", "Outside maintenance window — auto update deferred until the window opens", "Update anyway") appeared nowhere in the docs, so searching for the on-screen message returned nothing; Update Eligibility & Blockers now documents that dialog verbatim and states plainly that a soft blocker never gates a manual update.
Infrastructure-mode self-updates no longer fail to spawn the helper container on Docker Hub installs (#644). For containers labeled dd.update.mode=infrastructure, the self-update helper is spawned from Drydock's own image, but the reference was rebuilt as registry-1.docker.io/codeswhat/drydock:<tag> — the Docker daemon stores Hub images under their short name (codeswhat/drydock:<tag>), the helper image is never pulled since it must already exist locally, and container creation failed with 404 No such image, rolling the update back. The helper image reference is now resolved through the registry provider's getImageFullName normalization (the Hub provider strips registry-1.docker.io/, docker.io/, library/) so it matches the daemon-local image name; the previous raw construction remains only as a fallback when no registry manager resolves.

1.6.0-rc.9 — 2026-07-28

Added

Continuous Portwing edge log streams. The authenticated container-log WebSocket now bridges correlated dd:container_log_chunk, dd:container_log_end, and dd:container_log_error frames, cancels the agent stream when the viewer closes, preserves stdout/stderr and timestamp decoding, and caps each downstream viewer at 1 MiB of buffered data. Older Portwing agents degrade to their one-shot response without breaking the viewer.
Real Portwing fleet-soak workflow. Pull requests and the weekly quality schedule run actual signed Portwing processes against Drydock's production edge gateway through concurrent exec, sustained logs, forced controller backpressure, and reconnect storms. Machine-readable evidence records RSS/heap budgets and is retained for 90 days.

Changed

The portwing/1.0 edge endpoint is enabled by default. DD_EXPERIMENTAL_PORTWING=false remains as an emergency disable for new edge connections. OpenAPI, operator docs, and translated feature summaries now describe the stable/default-on behavior.

Fixed

The row "Updating/Queued/Scanning" overlay chip is no longer clipped at the pinned-column edge. The table's pinned identity-cluster cells (sticky z-10) painted over the full-width row overlay hosted in the first cell and cut the centered chip in half at the cluster boundary. While an overlay is active the host cell now sits above its z-10 row siblings (and still below the sticky header). (#631)
Registry errors no longer replace the container's tag in the tag column. A rate-limit/auth/not-found registry error rendered a danger pill instead of the current tag (table mode) or appended one (card mode), duplicating the warning glyph + tooltip the registry column already shows. The tag cell now always shows the actual tag; the registry column glyph remains the single error flag. (#631)

1.6.0-rc.8 — 2026-07-28

Fixed

Auto-update no longer stops when the update-available notification rule is scoped to specific channels (#623). Action triggers (docker, dockercompose, command) were gated by the same trigger allow-list on the update-available notification rule that routes messages to notification channels — but action-trigger ids are deliberately barred from that list by the API validator, the UI picker, and the documented rule model, so the moment any notification trigger was assigned to the rule, every action trigger (local and agent-hosted alike) silently failed the membership check with excluded-from-allow-list and auto-update stopped fleet-wide, with only a debug log as evidence. Action-category triggers are now exempt from the allow-list membership check in getUpdateAvailableAutoTriggerDispatchDecision (app/triggers/providers/Trigger.ts), mirroring the exemption the lifecycle-notification path has always had; disabling the rule itself still acts as the global kill switch. Present since the rule allow-list landed in v1.6.0-rc.1.
Controller-set update policy overrides no longer vanish from agent-managed containers (#565). Remote agents resolve their own declarative (env/label) policy but never learn controller-side runtime overrides, so every container report they send carries an explicit empty override layer. The controller persisted that layer verbatim, and updateContainer (app/store/container.ts) treated any present updatePolicyOverrides key as authoritative — clearing maturity mode/min-age days, skip lists, and snoozes on every periodic agent sync or manual recheck. This was the settings-deletion mechanism behind #565, distinct from the soak-clock resets fixed in #568 and rc.7. The controller now reapplies its stored overrides when ingesting agent reports, and the store itself encodes the rule the recreate path has had since #497: an empty incoming override layer carries no controller intent and only clears stored overrides when the update-policy PATCH handler marks the write as authoritative, so deliberate clears from the UI still stick.

1.6.0-rc.7 — 2026-07-26

Changed

Regenerated ui/src/boot/icon-bundle.json against the locked iconify versions (#608), fixing drift since the tabler 1.2.35 → 1.2.37 bump landed in #594.

Fixed

The maturity soak clock no longer resets when a container is recreated. getResultSignature (app/store/container.ts) decided whether a recreated container inherits its stashed updateDetectedAt / firstSeenAt / maturityGatePendingSince, and it included created even when a digest was present. Registries don't all derive created from the digest, so it could differ between the pre-delete snapshot and the post-recreate rescan without the update candidate changing — discarding the stash and silently restarting the soak right after an update landed, which is exactly when clock continuity matters. Same defect class as #565, surviving in the recreate path that #568 never touched.
Notification dedup no longer fires a duplicate once: true notification when a manual recheck bypasses the registry poll cache. computeResultHash (app/store/notification-history.ts) hashed suggestedTag and created alongside tag/digest/updateKind, so a recheck that returned a drifted created/suggestedTag for the same candidate produced a different hash, made hasAlreadyNotifiedForResult report "not yet notified", and re-fired notifications that were supposed to be one-shot. The hash now derives from the same candidate-identity definition PR #568 established for the maturity soak (tag + digest, with created participating only when no digest is available), via a single shared helper (getCandidateIdentityFields in app/model/container.ts) now used by hasCandidateIdentityChanged, getResultSignature, and computeResultHash alike — previously three independently hand-rolled comparators that disagreed with each other. Upgrading may produce one extra notification for updates already in flight, since history persisted under the old hash formula won't match the new one; this is a one-time false positive, not a regression.
A security scan finishing no longer reverts an update the watcher detected while it was running. runBulkScan (app/api/container/bulk-security.ts) and persistAndBroadcast (app/api/container/security.ts) both captured a container snapshot before starting the scan, then wrote that whole snapshot back with only security changed. Vulnerability scans run for seconds to minutes, so a watcher poll during that window could legitimately detect a new update and have result, updateAvailable, updateDetectedAt, and maturityGatePendingSince silently rolled back by the scan's write-back. The next poll re-detected the same update and stamped a fresh timestamp, discarding the elapsed soak time. Both paths now re-fetch the live record at write-back time and merge only security onto it, matching what the scan scheduler already did, and skip the write entirely if the container is gone.
Containers with no available update are no longer ranked or filtered as if they had one. getContainerUpdateAge (app/api/container/update-age.ts) fell back to firstSeenAt / publishedAt / updateDetectedAt without checking updateAvailable, so a container gated by maturityMode could still appear under ?maturity=mature, ?maturity=hot, and ?sort=age while its own updateEligibility blockers in the same response said it was blocked.

1.6.0-rc.6 — 2026-07-26

Added

Maturity-cleared notification (Discussion #587). When an update previously withheld by the maturity gate (maturityMode: mature) becomes applicable, drydock now fires a dedicated maturity-cleared notification instead of waiting for the next scheduled scan's generic update-available notification. A new background sweep (DD_MATURITY_SWEEP_CRON, default every 5 minutes) re-checks containers currently withheld by the gate and fires the moment each one's soak window elapses; the same check also runs at every watch-cycle scan. Delivery respects the same THRESHOLD, include/exclude, and ONCE dedup rules as update-available, and is deduplicated against it so the same update is never announced twice. Action triggers (Docker, Docker Compose, Command) are unaffected and keep applying updates on their normal scan cadence.

Changed

Routine dependency maintenance across every workspace. Minor and patch updates for the app, UI, demo, documentation website, and end-to-end suites, including undici 8.8.0, helmet 8.3.0, express-rate-limit 8.6.0, node-cron 4.6.0, re2js 2.8.6, and @aws-sdk/client-ecr 3.1091.0.
Translations resynced from Crowdin (#545), refreshing the container, dashboard, list, and shared-component catalogs across the supported locales.

Fixed

Empty-result fallbacks in CI scripts are reachable again (#599). Steps that run under bash -eo pipefail piped a grep or find straight into a variable assignment, so an empty result — no matching branch, a load-test artifact directory that a failed run never created — exited non-zero and aborted the step before the deliberate "nothing found" handler beneath it could run. The Crowdin sync hit this for real on 2026-07-24 and retargeted its translation PR at the default branch. The fallible command is now guarded on its own so a genuine failure (network, auth, bad syntax) still fails loudly, and regression tests pin every affected site.
Renovate no longer opens dependency PRs that bump package.json without the lockfile (#594), which previously landed manifests and lockfiles out of step. A lockfile-sync test now enforces the pairing.
The gitignored .planning directory is no longer tracked (#593), with a test guarding against re-adding files that .gitignore already excludes.

1.6.0-rc.5 — 2026-07-23

Changed

The Containers table's Resources column is now optional (#498). It remains visible by default, but the column picker can hide it and preserves that choice. Source, release-note, and registry shortcuts move into each row's More menu while hidden and remain available in card and detail views.
The Playwright release gate now waits for watcher-complete QA fixtures. Browser scenarios start only after the authenticated setup sees the full local and remote fixture set from the startup scans, while later cron and Docker-event scans are parked in the browser-only compose stack. This prevents grouping, update-status, and targeted-refresh assertions from racing partially enriched container state without adding whole-suite retries or changing production watcher defaults.

Fixed

The Dashboard and Containers views no longer overflow horizontally on narrow phone screens (#498). Crossing into a single-column breakpoint now re-syncs the dashboard widget layout so widgets stop spilling past the viewport, single-column card lists no longer leave an empty band below reflowed cards, and long stack names in the Containers group header truncate instead of pushing the update button off-screen.

Security

The documentation website now runs Next.js 16.2.11 and React 19.2.8. The Next.js patch closes nine upstream advisories disclosed against 16.2.9, including proxy bypass, Server Actions denial of service and SSRF, rewrite SSRF, response-cache confusion, image-optimization denial of service, and Server Function endpoint disclosure (GHSA-6gpp-xcg3-4w24, GHSA-m99w-x7hq-7vfj, GHSA-89xv-2m56-2m9x, GHSA-p9j2-gv94-2wf4, GHSA-68g3-v927-f742, GHSA-4633-3j49-mh5q, GHSA-4c39-4ccg-62r3, GHSA-q8wf-6r8g-63ch, GHSA-955p-x3mx-jcvp).
The documentation website's postcss build dependency is upgraded to 8.5.22, closing CVE-2026-45623 (GHSA-6g55-p6wh-862q), an arbitrary-file-read via an attacker-controlled sourceMappingURL (fixed upstream in 8.5.12).

1.6.0-rc.4 — 2026-07-22

Added

Container groups can be edited directly in the browser (#498). A container's More menu can set, change, or clear a local group override; clearing it falls back to the durable dd.group / Compose project / Swarm namespace group. Overrides are regular UI preferences, so they remain browser-local unless cross-device preference sync is enabled.

Changed

Same-tag rebuilds read “Image update” instead of “Digest update.” The update tooltip now explains that the visible tag has not changed but points to a different image build, and that redeploying pulls the new image. The dashboard uses the same vocabulary; literal image hashes are still correctly labeled digests.
The Containers table distinguishes Tag, Software Version, and Update more clearly. The secondary metadata column is now named Software Version, explains its source in a header tooltip, and folds before Host at constrained laptop widths so host identity stays visible.
The release-gated E2E lanes now fail with attributable evidence instead of blanket retries. Cucumber reuses the build-gated QA image, waits for the exact active fixture manifest, restores scenario-mutated state, publishes structured reports and diagnostics, and leaves browser rendering to Playwright. Playwright keeps first-failure media without retrying the whole test, gives its short-lived QA process explicit API and icon-proxy traffic budgets, waits for Docker's own healthcheck to acknowledge its health-transition fixture before scanning it, and fails setup if bounded runner-daemon pulls and image transfers cannot seed its required remote Docker fixtures. DD_SERVER_RATELIMIT_MAX can override the outer API's default 1,000-request/15-minute budget per rate-limit key; deployed defaults remain unchanged.

Fixed

Pinned containers with a visible newer version no longer say “Current.” A non-actionable updateInsight now renders as an informational Major/Minor/Patch state in table, card, and Update Status views while remaining ineligible for automatic or manual update actions.
A targeted container recheck no longer overwrites fresh Docker health. POST /api/v1/containers/:id/watch now scans and returns the live container object refreshed by the watcher instead of the stale store snapshot captured before refresh, preserving health-only transitions and their notification events.

1.6.0-rc.3 — 2026-07-21

Changed

Update-status vocabulary overhaul across the containers UI (#498, #556). "Digest" reads as "Digest update"; the NEW/MATURE freshness badges are gone (the detection age lives in the update-kind badge's tooltip instead); an unrecognized update kind now renders a neutral "Unknown" badge instead of nothing; and the bouncer's "Blocked" label is now "Security hold".
Pinned is a tag property again, not an update status. Pinned containers show a persistent pin glyph on the Tag cell whether or not an update exists, with the dd.tag.family remedy in its tooltip; up-to-date pinned rows read "Current" like everything else instead of a bare "Pinned" chip that only appeared when a newer family version existed. The container-detail Update Status panel follows the same rule: an insight-only container reads up to date there too, with the held-back newer tag relegated to an informational detail row.
The maturity panel keeps exactly one clock. The policy sentence names the clock the gate actually measures against ("Candidate published {date} — n more days until the minimum", with a detection-date fallback), the duplicate countdown collapsed into one "{countdown} · unlocks {date}" line, and the backend now exposes the resolved clock (clockSource/clockStartAt) on the maturity blocker so the UI reads the gate's own verdict instead of re-deriving it from the detection time.
The column picker tells the truth about auto-hidden columns. Columns hidden to fit the viewport stay checked with a muted "hidden to fit" note, and the "+N" toolbar badge's tooltip names them.

Fixed

The pin glyph marks only tags the pin gate actually governs. The glyph is driven by a new backend tagPinGated verdict — a specific-version tag with no dd.tag.include filter under a non-loose dd.tag.family policy — instead of bare tag shape, which had put a "won't climb to newer versions" pin (clipped to near-invisibility) on every exact-version container, including ones with actionable updates. It now renders inline beside the pinned tag at a legible size.
Container-detail registry links land on the registry they name (#556). The /registries?q=<type>.<name> deep-links emitted by container detail matched neither the bare registry name nor the bare type, so they always rendered an empty filtered list; the registries filter now also matches the dotted instance ID.
Removed-API-path banner grammar and tooltip freshness. The banner title pluralizes correctly ("1 request" vs "n requests"), and tooltip-bound native title attributes now follow reactive value changes (previously the column picker's "+N" badge kept naming whichever columns were hidden when the page first rendered).
Manual recheck no longer restarts the maturity countdown on display-only metadata drift (#565). A per-container recheck bypasses the registry poll cache, so the suggested tag and image created date can wobble between scans even when the update candidate itself hasn't changed, falsely resetting the soak. The lifecycle clock now restarts only when the candidate's actual identity — tag or digest — changes; suggested tag and created date no longer factor into the restart decision.

Security

Release dependency security refresh. Updated fast-uri to 3.1.4 for CVE-2026-16221, fast-xml-parser to 5.10.1 for GHSA-8r6m-32jq-jx6q, and the website's transitive sharp dependency to 0.35.3 for GHSA-f88m-g3jw-g9cj.
Anonymous access now fails closed on upgrades, not just fresh installs. An instance with no authentication configured — or with anonymous auth enabled but unconfirmed — starts and fails closed: protected API requests are rejected with 401, auth discovery/status remains public, /health reports 503, and the SPA shell may load without access to protected application data. This replaces the previous warning plus open dashboard. If you run an intentionally open instance, set DD_ANONYMOUS_AUTH_CONFIRM=true; otherwise configure DD_AUTH_BASIC_<name>_USER/_HASH before upgrading.
HTTP notification trigger hardened against SSRF. Hostnames are re-resolved on every request through a guarded DNS lookup that blocks cloud metadata and link-local targets (override with allowmetadata=true), and redirects can no longer escape into blocked address space via cross-host or DNS-rebinding hops.
WebSocket upgrades validate the complete origin. Origin checks now compare full scheme/host/port origins — honoring X-Forwarded-Proto/X-Forwarded-Host only when trust proxy is enabled — instead of host-substring matching, closing WebSocket CSRF gaps behind reverse proxies.
Session cookie renamed to drydock.sid. Replaces Express's default connect.sid, so drydock sessions no longer collide with (or are fingerprintable as) other Express apps behind the same host. Everyone is signed out once on upgrade.
Persistent store files are owner-only. /store is created 0700, dd.json is kept 0600, and the container entrypoint and store process set a 077 umask so LokiJS autosave replacement files never widen permissions.
Icon CDN sources are pinned to exact upstream revisions (dashboard-icons, selfh.st icons, simple-icons) instead of floating tags, so an upstream push can't silently change what the dashboard serves.

1.6.0-rc.2 — 2026-07-18

Added

"Version update" container filter (#538). The Containers filter bar gains a Version Update kind option that shows only real semver upgrades (major, minor, patch) and hides digest-only churn — the noise that dominates the dashboard on fleets that rebuild images daily in CI. Like every other container filter, it round-trips through the URL (?filterKind=version), so the filtered view can be saved as a bookmark for one-click access.

Fixed

Repair rebuilds no longer reset the registry-reconciled digest (#541). When the fast refresh path repairs a stored container's broken image reference, the reconciled digest.value — the security scan-group key — is now preserved unless the repo digest genuinely changed (image re-pulled), the same stickiness rule the slow-path rebuild gained for #536.
Containers with a stored watch error refresh on the fast path (#542). An error from the previous cycle no longer routes a known container through the full first-discovery enumeration — image/label/tag re-resolution plus an O(n) stale-entry scan of the store — on every cron. Errored-but-known containers reuse the same cheap refresh as healthy ones, and broken image references still self-heal via the unconditional repair check.
Last-known-good update state survives watch errors (#543). A failed watch cycle — registry timeout, rate limit, transient network error — no longer erases the previous successful comparison. The stored result, updateAvailable, updateKind, and current release notes are preserved alongside the recorded error until a later cycle succeeds, so the UI keeps showing the last known update state with the error annotated next to it instead of a blank.
Portwing WS Welcome frames now report the actual server version — the endpoint had hard-coded 1.5.0 since v1.5.0, so v1.5.1/v1.5.2/v1.6.0-rc.1 servers all identified as 1.5.0 to agents (#550 review finding)

1.6.0-rc.1 — 2026-07-15

Added

Actionable dry-run and update-preview UX (Discussion #321). Docker and Compose action triggers with DRYRUN=true are now marked in trigger cards, container actions, and the Update Status panel; the action is labeled Preview only, prevented replacements log at WARN, and History records update-applied-dryrun instead of claiming the container was replaced. Preview API failures now carry stable codes, sanitized details, and safe deep links for trigger/registry configuration, which the UI preserves in an accessible 44 px action.
Fail-closed no-worse-than-current security gating (Discussion #321). DD_SECURITY_GATE_RELATIVE=true keeps the configured absolute severity threshold, but permits an otherwise-blocked candidate when its CRITICAL, HIGH, MEDIUM, LOW, and UNKNOWN counts are each less than or equal to the exact running image's saved scan. Missing, failed, or stale current scans remain blocked. The persisted candidate scan, runtime API, OpenAPI contract, and audit row expose the comparison decision and evidence.
Rolling release-candidate image channel (Discussion #321). Canonical vX.Y.Z-rc.N cuts now publish X.Y-rc alongside the immutable exact tag on GHCR, Docker Hub, and Quay. Stable X.Y, X, and latest tags remain GA-only. Quick Start documents the tag matrix and recommends exact RC tags for reproducible reports.
Provider-neutral scanner runtime and off-heap SBOM lifecycle. Vulnerability scanning now supports trivy, grype, or normalized/deduplicated both mode across the compatibility command backend, hardened digest-pinned ephemeral Docker workers, and a Trivy-server remote composition that still models its client leg honestly. Scanner availability defaults fail-closed, with an explicit audited warn opt-in that never bypasses known blocking findings. The Security view and API expose provider/worker health plus audited pull/warm actions. Grype-only SBOM generation automatically uses Syft. New and migrated current/update SBOM bodies live in atomic checksum-validated, content-addressed /store/sbom blobs; LokiJS rows retain references and the API lazily dereferences them.
Global update mode and actionable Update Status panel (Discussion #325). Settings → General now selects one server-wide update mode: notify detects and notifies but refuses every Drydock-managed update, manual allows UI/API updates but suppresses automatic action-trigger dispatch, and auto allows both configured automatic actions and manual updates. The container side panel and full-page detail view replace the legacy eligibility-badge stack with one plain-language status plus a structured condition list covering all 16 eligibility reasons; actionable conditions open the relevant in-app policy, operation, security, or audit surface, or the corresponding configuration documentation when no editor exists. Maintenance-window deferrals are enriched into container-list and SSE status for local and agent-owned containers. Hard blockers disable the manual CTA, soft blockers retain the warn-and-confirm override, maturity/snooze conditions retain their lift time, and notifier-only mode collapses the eligibility detail behind an optional disclosure. Fresh installations default to manual; existing installations that predate the setting migrate to auto so an upgrade does not silently disable configured automatic updates.
Per-rule, per-trigger notification templates and bell preferences. The Notifications view can override simple title, simple body, and batch title independently for each notification rule and provider, render a live preview against representative event data, and save the overrides through the versioned notification API. The same rule editor now controls whether each supported event category appears in the in-app bell; update-available entries can be limited to major, minor-and-up, patch-and-up, or all updates. Existing trigger-level templates remain the fallback, so current delivery behavior is unchanged until an override is saved. (Discussion #205)
Zero-dependency custom dashboard grid. The dashboard no longer depends on grid-layout-plus. Its CSS Grid implementation packs layouts deterministically, supports edit-mode drag/reorder and bounded pointer resizing, preserves per-breakpoint layouts and hidden widgets, handles touch drag gestures, and keeps the existing reset/persistence behavior while fixing awkward same-row reorder interactions. (#281)
Declarative update policy with three-tier precedence. Docker containers can declare dd.updatePolicy.maturityMode, dd.updatePolicy.maturityMinAgeDays, dd.updatePolicy.skipTags, and dd.updatePolicy.skipDigests; Docker watchers can set maturity defaults with DD_WATCHER_<name>_MATURITY_MODE and _MATURITY_MIN_AGE_DAYS. Each field resolves watcher environment → container label → persistent UI/API override, with effective/declarative/override/source metadata exposed by the container API. The policy editor marks values that materially override their declaration and can revert one field or all fields without disturbing snoozeUntil. Overrides survive agent refreshes and container recreation, label removal remains authoritative beneath them, maturity blockers name their active source, and override set/clear operations are audited with every tier value. Motivated by Discussion #307 and tracked by #320.
Maturity stabilization countdown and override UX (Discussion #406). A candidate held by maturityMode: mature is visible immediately in container list, card, detail, and dashboard update surfaces instead of looking current or up to date. The maturity blocker shows a live minute-by-minute countdown plus the exact local date and time when the gate lifts. If a newer tag, digest, or mutable-tag image supersedes the waiting candidate, the soak clock and displayed ETA restart. Update Now remains available as an explicit soft-policy override, with a warning that names the maturity blocker; automatic enqueue paths continue to respect the gate. The displayed time is the eligibility/gate-lift time, not a guaranteed deployment time: automatic application still occurs on a subsequent watcher check and can be delayed by other blockers or maintenance windows.
Informational version visibility for pinned tags (#498). A specific-precision, unlabeled, non-loose pinned tag (e.g. nginx:1.25.3) still gets digest-only comparison for update actions — that pin-gate behavior from rc.2 is unchanged — but the container result now carries a new updateInsight: { tag, kind } field showing the best newer same-family tag that exists in the registry, purely as information. It reuses the exact same strict-style family matching used for actionable updates (prefix + suffix/variant compatibility + matching numeric-segment count + CalVer leading-zero rules) — no exception is carved out for major-version jumps; strict matching never restricted those to begin with. One narrow widening is scoped to this informational channel only: a prerelease-pinned tag (e.g. 1.5.2-rc.1) can see its own bare GA release (1.5.2) here, but that never makes the bare GA release an actionable update candidate — the actionable path rejects it just as before, even under dd.tag.family=loose or a permissive dd.tag.include filter. Ties at the same numeric version prefer the tag whose suffix template exactly matches the pinned tag's. This is additive only: updateAvailable, updateKind, and trigger dispatch are all unaffected. v1.6 adds the full policy chains — dd.tag.pin.info label → matching imgset → watcher TAG_PIN_INFO → true, plus watcher-level TAG_FAMILY as the lowest configured actionable default beneath labels and imgsets (strict remains the built-in default) — and replaces the hover-only badge with an at-a-glance informational treatment: grey current tag → blue newer tag, a neutral Pinned state, and a Major/Minor/Patch chip across list, card, and detail surfaces. The v2.7.5-openvino → v3.0.2 Major report in #498 was traced to an explicit dd.tag.family=loose label on that one Immich container, not the default pinned path; the historical bug that let loose mode cross from a suffixed variant to a bare tag is fixed and regression-covered, so variant comparisons remain variant-compatible (for example, -openvino stays -openvino).
Opt-in wud-card compatibility endpoints. The unversioned /api/* alias is removed in v1.6.0 (see the Removed section below and DEPRECATIONS.md), but the Home Assistant wud-card integration (and Homepage's native whatsupdocker widget, which shares the same contract) only speaks that unversioned, WUD-shaped surface. Setting DD_COMPAT_WUDCARD=true (default false) mounts a narrow, genuinely self-sufficient compatibility layer at /api/* — served by the compat router's own internal API router instance, not by falling through to the removed alias, so it keeps working now that alias is gone — covering exactly the four endpoints wud-card calls (GET /containers, GET /containers/:id/triggers, POST /containers/watch, POST /containers/:id/triggers/:triggerType/:triggerName). Three of those four are reshaped into WUD's bare-array response instead of drydock v1's { data, total, limit, offset, hasMore, _links } envelope; the trigger-run endpoint already returns a small non-enveloped body and passes through unmodified. Everything else under /api/* now returns 410 Gone. Off by default, subject to the same authentication and rate limiting as the rest of the API, and best-effort with no compatibility guarantee. (Discussion #469)
Experimental Portwing edge agents now actually serve container logs and deletes over the WS tunnel. EdgeAgentAdapter is wired into AgentClient's getContainerLogs()/deleteContainer() dispatch — a container hosted behind an edge agent (DD_EXPERIMENTAL_PORTWING=true) now routes log-tail and delete requests over the existing wss:// connection instead of falling through to a nonexistent HTTP endpoint on an edge-agent placeholder host. (#470)
Edge log/delete responses are correlated by echoed requestId, and the timestamps option is forwarded over the tunnel. A current Portwing agent echoes back the requestId drydock sends on dd:container_log_response / dd:container_delete_response, so EdgeAgentAdapter now resolves the exact originating request — correct even when two requests for the same container complete out of order — instead of the previous oldest-outstanding-by-containerId (FIFO) heuristic, which is retained only as a fallback for older agents that don't echo. Log downloads through an edge agent also honor the timestamps query parameter now (the wire message carries a timestamps field the agent reads), so the UI "show timestamps" toggle works over the edge path the same as for HTTP/SSE agents. Resolves the transport gaps previously documented as Bug 4 and punch-list #5.
Portwing WS connections are now proactively closed when an agent stops answering pings. The server sends a ping every 30s and expects a pong in return; an agent that misses two consecutive cycles (60s, matching portwing's own readDeadline) is treated as dead and the connection is force-closed, freeing the agent slot immediately instead of leaving a zombie connection registered until the underlying transport eventually notices. (#470)
Bidirectional MQTT: Home Assistant can now trigger updates back, not just observe them. Setting DD_NOTIFICATION_MQTT_{trigger_name}_HASS_COMMANDS=true adds a command_topic to each container's discovery payload, so the update entity's Install button in Home Assistant becomes clickable. Clicking it publishes to a per-container command topic that drydock subscribes to, which dispatches through the same update path the /update/:containerName webhook already uses, so all existing eligibility checks, trigger resolution, and agent routing apply unchanged. Commands are rate-limited per container (one accepted click every 30s) and every outcome (accepted, rejected, unexpected error) is recorded in the audit log under the new mqtt-command-update action. (Discussion #210)
Edge agents can choose their own display name via hello.agentName. Previously every edge agent was named portwing-edge-<agentId> unconditionally. A supplied name is now sanitized to a safe slug (lowercase, alphanumeric + hyphen, max 63 chars) and used as the registry/display name, falling back to the old portwing-edge-<agentId> form when absent or empty. The name is bound to the authenticating Ed25519 key on first use — a later hello reusing the same name is only admitted under the same key (rejected with an agent-name-claimed error otherwise), and the binding is released when its owning key is revoked — so one registered key can no longer squat or steal another agent's chosen name. The binding is persisted (a name-bindings store collection, reloaded into memory on startup) so this guarantee holds across a server restart too, not just for the lifetime of one process. (#470)
Opt-in Ed25519 request signing for standard-mode agents. A controller agent's config gains an authmode setting (token | ed25519, default token, so existing configs are unchanged). Setting DD_AGENT_{name}_AUTHMODE=ed25519 — plus DD_AGENT_{name}_SIGNINGKEYID and a PEM PKCS#8 DD_AGENT_{name}_SIGNINGKEY (or _SIGNINGKEY__FILE) — makes the controller sign every request to that agent with an Ed25519 key instead of sending the shared X-Dd-Agent-Secret. Each request carries four X-Portwing-* headers (Key-ID, Timestamp, Nonce, and a base64url Signature) over the canonical METHOD\nPATH\nSHA-256-hex(body)\ntimestamp\nnonce message, matching Portwing's verifier, so a captured signature can't be replayed and no long-lived secret crosses the wire. The signing key is parsed and validated once at startup (fail-fast on a malformed key), masked in configuration dumps like the secret token, and a missing key id/key skips just that agent with a warning rather than crashing the controller. Because an ed25519 agent transmits no secret, it connects over plain HTTP without DD_AGENT_ALLOW_INSECURE_SECRET. (#470)
Cross-device preference sync. An opt-in Sync across devices toggle in Config > Appearance (off by default, hidden for anonymous sessions) stores the full UI preference set — theme, layout, dashboard, language, and more — server-side per user via a new GET/PATCH /api/v1/preferences endpoint, and propagates changes to a user's other signed-in devices in real time over the existing SSE connection. Turning sync off keeps the server-side copy but stops syncing, and the device's own localStorage becomes authoritative again. (Discussion #220)
Health-status event notifications. A new container-unhealthy notification rule fires your existing notification triggers (Slack, SMTP, webhook, etc.) the moment a Docker health check enters the unhealthy state — detected on both the per-event Docker health_status listener (near-instant) and the 6-hour cron fallback, for a container that has previously been observed in a non-unhealthy state within the same instantiation. Never fires for a container with no HEALTHCHECK configured, for a container discovered already unhealthy with no prior baseline, or repeatedly while a container stays continuously unhealthy — but does fire again after a fast restart/crash-loop resets the container's boot timestamp, even if the health status reads unhealthy on both observations. Disabled by default, matching the agent-disconnect/agent-reconnect precedent; enable the container-unhealthy rule and assign triggers (or leave the trigger list empty to fire to every notification trigger) to receive it. No recovery/"healthy again" companion event, no dd.autoheal* label, and no corrective action ship with this — the full auto-heal loop stays a later, separately-scoped feature. (Discussion #198)
Column show/hide picker now covers every major data view. Audit, Security, Servers, and Watchers gain the same per-view column show/hide picker (DataTableColumnPicker, backed by the useViewColumnVisibility composable) that Containers and Agents already had, so every one of those six views can hide low-priority columns and have the choice persist per view.

Changed

Registry requests now retry transient network failures. Response-less network errors — timeouts (ECONNABORTED/ETIMEDOUT), connection resets (ECONNRESET), and temporary DNS failures (EAI_AGAIN) — are retried up to twice with exponential backoff before a watch error is recorded, complementing the existing 429/503 Retry-After handling. A brief registry blip no longer marks affected containers as errored until the next hourly cycle.
Legacy trigger-taxonomy inputs now emit error-level migration signals in their final compatibility release. Every detected DD_TRIGGER_* variable and deprecated dd.trigger.include / dd.trigger.exclude label remains functional in v1.6, but is logged at error level and remains scheduled for removal in v1.7. Use DD_ACTION_* / DD_NOTIFICATION_* and category-scoped labels, or run config migrate --source trigger.
Registry polling and hot container summaries do less duplicate work. Tag-list requests for the same registry repository are now shared across containers within a poll, including concurrent lookups, without leaking mutable cached arrays to callers. Dashboard summary uses the lightweight stats projection, and security overview reads the collection without deep-cloning every container before building its aggregate response.
Scheduled security scans no longer flood History with unchanged container updates. Security results still refresh the container store, UI, and notification paths, but the generic container-update audit row is now emitted only when meaningful lifecycle, image, update, error, health, or policy state changes. Transition state uses the persisted compose-aware container identity, stays isolated across agents, watchers, and same-name Compose siblings, and is cleared when that exact container is removed.
The shared application log viewer stays responsive with large histories. JSON highlighting now uses the existing bounded tokenizer cache, and collections above 200 entries render through a measured, overscanned virtual window while preserving search navigation, wrapping, line numbers, auto-scroll, and live row-height changes.
CI dependency metadata is cleaner. The Playwright workflow comment now matches its actual Chromium-only matrix (PR #486), and the unused @types/node-cron package was removed from the backend workspace. Both fixes are applied directly on dev/v1.6; the Renovate PRs remain open until default-branch convergence.
Base image bumped to Alpine 3.24. The healthcheck-build stage in the Dockerfile moves from alpine:3.21 to alpine:3.24.
Every list view toggles between table and cards, and the choice sticks per view. The v1.6 UI refactor rebuilt all list views on a shared DataTable, and each filter bar's view switch is now table⇄cards across Containers, Agents, Notifications, Security, Triggers, Watchers, Servers, Registries, Audit, and Auth. The selected mode is persisted per view; below ~640px the layout automatically reflows to cards and the now-redundant toggle is hidden. The old three-way list (accordion) mode has been removed — it's table or cards.
Card view gained a sort control. Containers, Agents, and Security now render a sort-field and ascending/descending control (DataSortControl) in the filter bar when in card mode — the pre-v1.6 card grid (DataCardGrid, since removed) had no sorting of its own, so cards showed whatever order the table view happened to produce.
Container resource shortcuts are now consistent across views (Discussion #295). The existing source-project and release-note capabilities now render through one Source → Release notes → Registry toolbar in container table/cards/details, Security table/cards/detail, and Dashboard Recent Updates. The Containers table has a required Resources column separate from lifecycle Actions; cards and compact layouts wrap resources onto their own row. Every resource target is 44×44 px, and registry opens the filtered internal Registries view instead of a raw OCI API endpoint. Release-note dialogs stay inside the viewport, contain touch scrolling and keyboard focus, reset when the selected row changes, allow only one open dialog, and restore focus when dismissed with Escape or Close. The removed accordion list mode is not a current target.
dd.action.* and dd.notification.* trigger labels are now strictly category-scoped. A container's dd.action.include/dd.action.exclude labels now gate only action triggers (docker, dockercompose, command); dd.notification.include/dd.notification.exclude gate only notification triggers. Previously, whichever of the two was set on a container silently won for both categories, so a lone dd.action.include also filtered notification triggers (and vice versa) — see the Fixed entry below. The deprecated dd.trigger.include/dd.trigger.exclude labels still apply to both categories as a shared fallback beneath the scoped labels, unchanged. See the Upgrade Notes entry below before upgrading if you rely on a single scoped label to gate both trigger categories.

Deprecated

GET /api/auth/methods. This unversioned auth-discovery alias now logs on every request, returns Deprecation/Sunset headers, and points callers directly to canonical GET /api/v1/auth/status. It is documented in DEPRECATIONS.md — deprecated in v1.6.0, removed in v1.7.0.
Legacy auth strategies response shape (GET /auth/strategies). This endpoint's older { strategies, warnings } response shape is superseded by GET /api/v1/auth/status's { providers, errors } shape. It now logs on every request and returns Deprecation/Sunset headers; removal is scheduled for v1.8.0.

Removed

Legacy v1.4-era authentication compatibility. Basic authentication now accepts only argon2id hashes; {SHA}, APR1/MD5, crypt, and plain-text hashes fail validation. OIDC discovery now requires https://; HTTP discovery no longer enables an insecure client workaround.
Legacy WUD configuration runtime aliases. WUD_* environment variables and wud.* Docker labels are ignored. The migration CLI intentionally still recognizes them so existing files can be rewritten to DD_* and dd.* before startup.
Obsolete Docker watcher switches. DD_WATCHER_<name>_WATCHDIGEST and WATCHATSTART are no longer configuration keys. Use dd.watch.digest=true per container; startup watches are always scheduled.
Legacy trigger-template aliases. $id, $name, $watcher, $kind, $semver, $local, $remote, $link, and $count are no longer populated. Templates use the canonical container/update context and $containers.length.
Kafka clientId and token-only public-registry compatibility handling. Kafka validation accepts only lowercase clientid; malformed Hub/DHI public instances such as PUBLIC_TOKEN without PUBLIC_LOGIN now fail closed instead of silently falling back to anonymous pulls. Valid LOGIN+TOKEN, LOGIN+PASSWORD, and AUTH configurations remain supported.
Unversioned /api/* alias removed. Deprecated since v1.4.0, the unversioned /api/* prefix now returns 410 Gone with a JSON body pointing callers at the /api/v1/ equivalent instead of serving the request. This is a breaking change for the Home Assistant wud-card integration and Homepage's native whatsupdocker widget — both hardcode the unversioned /api/* base path with no way to configure a different one, so they start getting 410s the moment they hit drydock v1.6.0. Their supported path forward is DD_COMPAT_WUDCARD=true (default false), which mounts a narrow, self-sufficient compatibility layer serving exactly the four endpoints those integrations call (see the Added section above and DEPRECATIONS.md). Everyone else migrates to /api/v1/*. GET /api/auth/methods is not part of this removal — it's mounted directly on the app, ahead of the /api/* mounts, and stays on its own separate deprecation schedule (removed in v1.7.0).
Unversioned WebSocket alias WS /api/log/stream removed. Same removal as the REST /api/* alias above, applied to the log-stream upgrade: connecting to the unversioned path now fails fast with a 410-style upgrade rejection instead of completing the WebSocket handshake. Use the canonical WS /api/v1/log/stream endpoint.

Fixed

Recovering from a registry outage no longer corrupts a container's stored image identity. When a container's stored record carries a watch error, the next watcher cycle rebuilds it from scratch — and that rebuild blindly reset image.digest.value to the raw local image digest, discarding the registry-reconciled manifest digest recorded on the last successful cycle. The reset flipped the security scheduler's scan-group key, forcing a needless fresh vulnerability scan and emitting a spurious container-update audit row for every container sharing the image. Surfaced by the v1.6 24-hour store-size acceptance run, where ten such rows during a Docker Hub outage were exactly the audit-growth margin of failure (Discussion #321). The rebuild now preserves the reconciled digest value unless the local repo digest genuinely changed (a real re-pull), matching the clean-path refresh semantics.
User-set update-policy overrides survive error-state rebuilds. The same error-state rebuild stamped updatePolicyOverrides to {} on the rebuilt record, silently destroying stored snoozes and skipped tags/digests for any container that had a watch error. Overrides are now seeded from the existing stored record before declarative policy is re-applied.
Per-container update policy survives Drydock-triggered updates (#535). During an update, Drydock renames the outgoing container to a transient <name>-old-<timestamp> rollback alias; the Docker rename event wrote that alias onto the still-tracked container record (its display name too, when no custom dd.display.name label is set). The poisoned name then defeated the v1.5.2 policy-retention safeguard: post-update pruning could no longer match the record by name, and the rollback-alias guard skipped the policy hand-off, so the replacement container came up with no active update policy. The rename event now recognizes Drydock's own transient rollback alias — provable because stripping the alias suffix reconstructs the record's current name — and ignores it, while genuine renames (including containers legitimately named with an -old-<digits> suffix) propagate unchanged.
Failed container recreates now reclaim orphaned replacements before rollback. If creation succeeded but a later network-connect or start step failed, every update, self-update, health-monitor, backup-restore, and Compose rollback path now recovers and removes the partial replacement before restoring the original name. Containers already stranded under nested -old-<timestamp> names are blocked from another cascading recreate with a manual-cleanup error. Recreates also stop re-pinning daemon-assigned MAC addresses, while preserving explicitly configured primary-network MACs. Locally built images skip registry lookups, and unprefixed Docker Hub throttle warnings identify docker.io. This patch also ships on the consolidated v1.5.2 line. (PR #503)
Store growth is attributable and repeated update audits are bounded (Discussion #321). Diagnostic dumps now report UTF-8 serialized bytes for every LokiJS collection and the store total. Identical update-available reports are deduplicated by agent, watcher, container, and version transition for a configurable window (DD_AUDIT_UPDATE_AVAILABLE_DEDUPE_MS, one hour by default); target changes and no→yes transitions are recorded immediately.
Large-image Trivy scans no longer race their own timeout or discard the pulled candidate on scanner errors. The default DD_SECURITY_TRIVY_TIMEOUT is now 10 minutes, while Node gives Trivy an additional 30-second process grace so Trivy can report its own deadline instead of being killed as exit=unknown. The update gate retries one classified transient scanner failure, retains the pulled image when scanning itself errors, and still prunes images that are genuinely blocked by vulnerability policy. Local Trivy mode performs a serialized, single-flight --download-db-only warm-up outside the scan command's budget; server mode skips local warm-up. Scanner failures remain fail-closed. (#490)
The bundled Trivy binary now comes from an immutable official multi-architecture image digest. The release image, default Docker scanner worker, and release-cut SBOM job use the same pinned Trivy release instead of installing a floating Alpine edge package. Cosign remains version-pinned from Alpine 3.24; curl remains in v1.6 for compatibility with user-defined health checks and is scheduled for removal in v1.7.
Image builds no longer fail on a stale Alpine tzdata pin. Alpine 3.21's repositories replaced tzdata 2026b-r0 with 2026c-r0, which made every image build fail at apk add with an unsatisfiable exact pin. The base-image pin now installs the available 2026c-r0, and a regression test asserts the current revision is pinned (and the stale one absent) so a future rotation fails fast in CI instead of at release time. (PR #523; applied to main after the v1.5.2 cut as PR #533)
Registry and runtime hardening is consistent across provider-specific paths. Credential refresh, custom TLS settings, redirect handling, pagination cursors, and Docker Hub metadata requests now use the same bounded/fail-closed rules across supported registries. Secret-file loading, hook command policy, template property access, proxy-aware throttling, and API error responses were hardened in the same review pass.
Agent reconnect and security-digest state are bounded and lifecycle-safe. Removing an agent can no longer leave a reconnect queued, edge reconnect notifications reflect the real reconnect state, and digest notification buffers now expire and enforce configured limits across active and restored state.
A stalled auth bootstrap request no longer leaves the app blank indefinitely. The /auth/user request times out after eight seconds and falls through to the existing logged-out redirect path.
Authenticated session checks no longer trip the shared auth rate limiter. GET /auth/user calls from an already-authenticated session are now exempt from the shared public auth limiter (100 requests/15 min); normal signed-in navigation could exhaust that budget through repeated re-reads and start returning spurious 429s even though the backend was healthy. Every other auth route — including unauthenticated /auth/user checks, status discovery, login, remember, logout, and mutations — remains rate-limited as before. (PR #526)
Old preference schemas no longer skip intermediate migrations. Schema-v3 data now advances through each numbered migration, so later additions such as the softwareVersion column are applied before reaching the current schema.
Open tabs recover from stale lazy-loaded chunks after an upgrade. Vite preload errors and matching Vue Router dynamic-import failures request one guarded page reload; successful navigation clears the session guard so a future upgrade can recover independently.
Repeated stale-chunk failures are no longer silently swallowed after the guarded recovery attempt is exhausted. The first matching preload failure still requests one session-guarded reload; later failures continue through Vite's normal error path so monitoring and the host page can surface them.
Notification-bell controls now match their audit-backed event coverage. container-unhealthy joins the bell query, rules without a corresponding bell audit action (currently agent-reconnect) no longer show controls that could not take effect, and health-only changes now propagate through local and agent lifecycle events. Unhealthy audit dedupe is scoped by agent + watcher + container, and a replayable health-transition SSE is ordered after audit processing so the bell refetch sees any newly inserted row. Severity thresholds remain scoped to update-available; notification-delivery failures remain always visible.
Digest-only updates remain visible under every notification-bell severity setting. Update audit entries now retain whether the change is tag- or digest-based, so major, minor, and patch settings no longer discard digest changes whose semantic-version severity is necessarily unknown. Unrelated unknown-severity tag entries remain filtered as before.
The live system-log viewer keeps advancing after its browser buffer fills. Rolling past the 2,000-entry client cap now replaces the reactive array instead of mutating it in place, so newest-first sorting and virtualization observe every rollover rather than freezing the visible newest row while the WebSocket continues receiving entries.
Virtualized log position is stable while reading older entries. Measured row-height changes and newest-first batch prepends compensate both the virtual window and DOM scroll position; a screen-reader-only status also reports rendered versus total lines without reintroducing thousands of DOM nodes.
dd.tag.family=loose no longer bypasses the suffix/variant guard in isSemverFamilyMatch(). A pinned nginx:1.2.3-ls132 container could previously be offered a bare 1.2.4 (wrong variant) as an update candidate under loose policy — loose mode was only ever meant to relax prefix equality and CalVer leading-zero rules, not let updates cross a suffix/variant boundary entirely. The guard now applies unconditionally regardless of policy.
The candidate sort in sortSemverDescending() now prefers the exact-suffix-template match over a merely-compatible one when two candidates tie at the same numeric version (e.g. preferring 1.2.5-alpine over 1.2.5-alpine3.21 for a 1.2.3-alpine reference). Semver treats the suffix as a prerelease field, so without this fix the wrong variant could outrank the exact match purely on prerelease-string ordering.
Pinned semver tags are now compared by digest by default, as originally announced in v1.5.0-rc.36. A rebuilt image republished under the same tag is detected again. Previously digest watching was silently disabled for fully-pinned tags (e.g. nginx:1.25.3), leaving them with no update detection at all and a misleading "compared by digest only" notice on every pinned container. Version climbing for pinned tags remains opt-in via dd.tag.include or dd.tag.family=loose. In agent deployments, agents perform the registry checks — update agents alongside the controller to restore digest detection for agent-watched containers. (#498)
The "no update detection" notice is honest when digest watching is explicitly disabled. When dd.watch.digest=false (or an imgset watch.digest=false) is set on a pinned or floating tag, the notice no longer claims a digest comparison is happening. (#498)
Removed the unreachable flat tagFamily imgset config key. Env-derived config keys are lowercased, so the camelCase key could never match any real configuration; the documented DD_WATCHER_{watcher}_IMGSET_{name}_TAG_FAMILY form is unaffected. (#498)
GET /api/v1/containers/backups (list all backups) is now reachable. The route was previously registered as router.get('/', getBackups) in the backup router, which was mounted at /containers after the container router — so the container router's own GET / handler shadowed it and the list-backups endpoint was never reachable. The route is now router.get('/backups', getBackups) and the backup router is mounted before the container router, making GET /api/v1/containers/backups accessible. The per-container backup endpoints (GET /api/v1/containers/:id/backups) were not affected.
App favicon now matches the refreshed website branding. The v1.5.1 brand refresh (#439) updated the website to the cropped whale "headshot" icon but left the in-app tab icon, Apple touch icon, and PWA manifest icons on the old full-body whale. The app now ships the same icon set as the website. The stale favicon.svg — which modern browsers preferred over the PNGs, so it kept showing the old mark — was removed, and the icon links carry a ?v=2 cache-buster so existing installs re-fetch instead of serving the aggressively cached old icon. (#439)
Deprecation lifecycle, warnings, banners, and docs now match the v1.6 runtime. The audit removed retired OIDC/hash banners, narrowed the consolidated UI banner to active v1.7 trigger-prefix inputs, corrected stale CORS/stats dates, documented the permanent auth-status alias, deferred PUT /api/v1/settings to API v2, fixed the auth migration target and Sunset metadata, added request-level signals to the legacy auth-strategies response, and moved every completed v1.6 item out of the active schedule. The published docs now distinguish removed runtime aliases from the migration CLI, which intentionally retains knowledge of old names solely to rewrite files.
A stale cached page no longer white-screens after an upgrade. The UI's SPA history-mode fallback used to return index.html (an HTML 200) for any unmatched path, including a request for a content-hashed /assets/*.js bundle that a previous version referenced but the upgrade deleted. Browsers refuse to run a module script served as text/html, so a page cached from the old build (typically by a reverse proxy that ignores the HTML's Cache-Control: no-store) painted nothing, and caching proxies could poison the asset URL with that HTML. Requests under /assets/ that miss now return a clean 404 (with no-store) instead of the shell, so a stale page fails fast rather than blank. A new Reverse proxy caching docs section covers the browser/proxy cache settings that avoid this. This does not require a stale entry-script to recover on its own — clearing the browser/proxy cache is still the fix for an already-blank page — but it stops drydock from turning a missing bundle into a silent white screen. (#466)
Responsive containers table: sticky columns and auto-hide column budget corrected. The auto-hide column budget now measures the table's real available width (ResizeObserver + content-box measurement) instead of an estimate that ran ~23px too generous whenever the detail panel was open, so columns no longer hide too aggressively or too little. The icon and name columns are now pinned together as a single sticky-left cluster so the icon can no longer scroll out from under the name column, the sticky separator border only appears on genuine horizontal overflow, and the sticky actions column no longer paints on top of the last data column when a table legitimately overflows.
Icon column no longer clips container icons. The icon column was 40px wide with 20px of padding, leaving only ~20px of content box for the 32px container icon — about 11.9px of every icon silently hung past the cell edge. The column is now sized (40→56px) to actually fit the icon.
Edge agent memoryGb was reported in decimal GB, not GiB. EdgeAgentAdapter divided the reported byte count by 1e9; drydock's own convention (and portwing's canonical MemoryTotalGB()) is binary GiB (1024³). An 8 GiB host previously showed as ~8.59 in the UI and API; it now reports the correct ~8.00.
Portwing drydockCompat mismatches now warn in both directions. The server previously only logged a warning when a connecting agent's compat major version was newer than the server's; an older agent connecting to a newer server produced no signal at all. Any major-version mismatch, in either direction, now logs a warning pointing operators at the compat matrix — the connection is still accepted either way.
Malformed hello.agentName on the portwing edge WS could crash the drydock process. The hello handler called .trim() on hello.agentName with no type check; a number, boolean, array, or object in that field threw a TypeError that surfaced as an unhandled promise rejection capable of taking down the whole process. The field is now validated for type and length before use — a malformed value closes just that connection with an invalid-agent-name error frame instead.
A reconnecting edge agent could be evicted by its own stale connection. The server-side ping-liveness check's forced close ran the same disconnect cleanup as a real WebSocket close event, but the underlying close/error listeners were never detached — so when the transport eventually noticed the connection was already gone, the disconnect cleanup ran a second time. Since agent removal matched by name only, that delayed second cleanup could evict a different, newly-reconnected agent sharing the same identity-derived name. Disconnect handling is now idempotent, listeners are detached before a forced close, and agent removal is instance-checked (only fires if the registry still holds that same connection instance under the name).
Concurrent log or delete requests for the same container over the edge tunnel no longer clobber each other. Pending edge requests were keyed by container id alone, so a second in-flight request for the same container overwrote the first's entry; the first request's timeout then deleted the second's entry, silently dropping the real response and leaving the caller with a spurious timeout. Requests now carry a unique per-call id embedded in both the pending-request key and the outgoing wire frame, correlated on response via a per-container FIFO queue.
dd.action.include/dd.action.exclude no longer leak into notification trigger filtering (and vice versa). Label resolution collapsed dd.action.* and dd.notification.* into a single container.triggerInclude/triggerExclude field at parse time, first-match-wins — so a container with only dd.action.include set had that value applied to every trigger regardless of category, silently gating notification triggers the label was never meant to touch. The two categories now resolve into independent actionTriggerInclude/actionTriggerExclude/notificationTriggerInclude/notificationTriggerExclude fields, and trigger matching reads the field scoped to its own category. The deprecated triggerInclude/triggerExclude fields are still populated (for /api/v1 readers, the persisted store, and mixed-version agents) but are no longer read by matching code. (#494, discussion #493)
HASS latest_version template no longer reports "Unknown" when no update is pending. The Home Assistant MQTT discovery latest_version template rendered an empty string for any container with no pending update (no result in the flattened state payload), which HA silently discards — and because HA blanks the whole update.* entity state when either version is missing, the entity read "Unknown" forever instead of resolving to "up to date". The template now falls back to the installed tag (image_tag_value) in both the digest and tag branches, and guards the digest slice so a digest-kind report carrying no digest no longer trips HA's Undefined[:15] slice error. (#491)
Home Assistant MQTT discovery entities now respect the same per-container trigger gating as state publishes. Discovery entity creation used to run unconditionally off container-added/updated events, independent of the mustTrigger() gating (rollback/agent scoping and the per-category dd.notification.include/dd.notification.exclude filters) that already governed the per-container state publish — so a container excluded from an mqtt trigger still got a Home Assistant entity whose state topic never received a message, leaving a permanent "Unknown" ghost entity in HA. Entity creation, state publishes, and #210 Install commands now all honor the same gate: a previously-created entity for a container that becomes excluded is cleaned up (its discovery config removed) the first time the container is seen, and an Install command addressed to an excluded container is ignored. (#491)
Container update policy is no longer lost when a container is recreated (#496). Container documents are keyed by Docker's container ID, which changes every time a container is recreated (image pull, docker compose up -d, or an update trigger firing). The replacement was stored as a brand-new document and the per-container update policy — maturity gate, skipped tags/digests, snooze — was silently dropped along with the old one. Because an absent policy means "no gating" rather than "default gating", affected containers then updated immediately instead of respecting their maturity soak. The policy now survives a recreate, for containers watched locally and through a remote agent alike. Present since 1.4.1, when per-container maturity policies were introduced.
The remote-agent prune path now distinguishes a recreated container from a removed one, so a container that reappears under a new ID keeps its update policy and its Home Assistant state topic, while a genuinely deleted container still has its discovery topics cleaned up.

Upgrade Notes

Existing installs preserve automatic updates; fresh installs start in manual mode. The new global updateMode setting defaults to manual only when Drydock creates a settings record for the first time. An existing settings record with no updateMode is migrated to auto, preserving the pre-v1.6 behavior in which configured action triggers could apply updates automatically. Review Settings → General → Update mode after upgrading if you prefer notifier-only or manual-only operation.
A lone dd.action.include/dd.action.exclude (or dd.notification.include/dd.notification.exclude) label stops filtering the other trigger category. Before this release, setting only dd.action.include on a container also filtered notification triggers as a side effect of the two labels collapsing into one internal field (#494). As of v1.6, each label filters only its own category. If you relied on the cross-category leak, previously-suppressed notification (or action) triggers may fire once on the first scan after upgrading. drydock logs a one-time warning per affected container naming the missing label; set the matching dd.notification.include/dd.notification.exclude (or dd.action.include/dd.action.exclude) to the same value to restore the previous filtering.

Documentation

Podman and Docker socket security docs refreshed. Added the v1.5.2 guidance for Podman's Docker-compatible API path (#152), clarified direct socket vs proxy/TCP behavior, documented remote TLS/OIDC watcher auth, and tightened the Docker socket security/FAQ/security-guide cross-links without claiming native Podman support has shipped.
Documentation parity audits completed. Reconciled the current configuration/env-var reference tables against runtime schemas and refreshed the API/OpenAPI/docs contract for the canonical /api/v1 surface. The API docs now cover the recently added authentication component endpoints, registry webhook signature auth, notification outbox actions, bulk container updates, backup listing, auth status aliases, and container summary/update-response details; the OpenAPI spec now matches the route security modes and response codes for those surfaces.

1.5.2 — 2026-07-13

Consolidates the 1.5.2-rc.1 … 1.5.2-rc.5 prereleases. Users upgrading from 1.5.1 get everything below; users already on 1.5.2-rc.5 receive no additional runtime changes.

Added

Informational version visibility for pinned tags (#498). A specific-precision, unlabeled, non-loose pinned tag (e.g. nginx:1.25.3) still gets digest-only comparison for update actions — that pin-gate behavior from rc.2 is unchanged — but the container result now carries a new updateInsight: { tag, kind } field showing the best newer same-family tag that exists in the registry, purely as information. It reuses the exact same strict-style family matching used for actionable updates (prefix + suffix/variant compatibility + matching numeric-segment count + CalVer leading-zero rules) — no exception is carved out for major-version jumps; strict matching never restricted those to begin with. One narrow widening is scoped to this informational channel only: a prerelease-pinned tag (e.g. 1.5.2-rc.1) can see its own bare GA release (1.5.2) here, but that never makes the bare GA release an actionable update candidate — the actionable path rejects it just as before, even under dd.tag.family=loose or a permissive dd.tag.include filter. What's new is exposing this comparison at all for pinned tags, since the pin gate itself already blocks acting on any of it. Ties at the same numeric version prefer the tag whose suffix template exactly matches the pinned tag's. This is additive only: updateAvailable, updateKind, and trigger dispatch are all unaffected, so nothing new fires because of it. On by default; opt out per watcher with DD_WATCHER_{name}_TAG_PIN_INFO=false. Surfaced in the UI as an informational "Newer available" badge next to the existing update-state badges.

Changed

Docs: documented the socket-proxy requirements for recreating containers with a static IP or MAC address (macvlan, custom networks). Tecnativa/docker-socket-proxy needs both POST=1 and NETWORKS=1 for containers on more than one network (the extra POST /networks/{id}/connect calls); sockguard needs request_body.network.allow_endpoint_config: true to permit endpoint configs that specify a static IP or MAC address — and as of sockguard v1.5.0 and later (forthcoming at the time of this release), that requirement also applies to POST /containers/create, so single-network macvlan/static-IP containers need the policy set too, not just multi-network ones (older sockguard versions enforce the policy only at network-connect time). Added a matching FAQ entry for containers left renamed -old-<timestamp> after a failed update, including a note for operators who intentionally name a container in a way that collides with drydock's own rollback naming convention.

Fixed

Container update policy is no longer lost when a container is recreated (#496). Container documents are keyed by Docker's container ID, which changes every time a container is recreated (image pull, docker compose up -d, or an update trigger firing). The replacement was stored as a brand-new document and the per-container update policy — maturity gate, skipped tags/digests, snooze — was silently dropped along with the old one. Because an absent policy means "no gating" rather than "default gating", affected containers then updated immediately instead of respecting their maturity soak. The policy now survives a recreate, for containers watched locally and through a remote agent alike. Present since 1.4.1, when per-container maturity policies were introduced.
The remote-agent prune path now distinguishes a recreated container from a removed one, so a container that reappears under a new ID keeps its update policy and its Home Assistant state topic, while a genuinely deleted container still has its discovery topics cleaned up.
Pinned semver tags (e.g. nginx:1.25.3) are now compared by digest by default, as originally announced in v1.5.0-rc.36 — a rebuilt image republished under the same tag is detected again. Previously digest watching was silently disabled for fully-pinned tags, leaving them with no update detection at all and a misleading "compared by digest only" notice on every pinned container (#498). Version climbing for pinned tags remains opt-in via dd.tag.include or dd.tag.family=loose. Note: in agent deployments, agents perform the registry checks — update agents alongside the controller to restore digest detection for agent-watched containers.
The "no update detection" notice shown when dd.watch.digest=false is explicitly set no longer claims digest comparison is happening.
Removed the unreachable flat tagFamily imgset config key (env-derived keys are lowercased, so it could never match); the documented DD_WATCHER_{watcher}_IMGSET_{name}_TAG_FAMILY form is unaffected.
dd.tag.family=loose no longer bypasses the suffix/variant guard in isSemverFamilyMatch(). A pinned nginx:1.2.3-ls132 container could previously be offered a bare 1.2.4 (wrong variant) as an update candidate under loose policy — loose mode was only ever meant to relax prefix equality and CalVer leading-zero rules, not let updates cross a suffix/variant boundary entirely. The guard now applies unconditionally regardless of policy.
The candidate sort in sortSemverDescending() now prefers the exact-suffix-template match over a merely-compatible one when two candidates tie at the same numeric version (e.g. preferring 1.2.5-alpine over 1.2.5-alpine3.21 for a 1.2.3-alpine reference). Semver treats the suffix as a prerelease field, so without this fix the wrong variant could outrank the exact match purely on prerelease-string ordering.
Tooltip text now wraps inside a bounded popup instead of rendering one screen-wide line — the shared tooltip directive had white-space: nowrap and no max-width, so the pinned-tag "Newer available" insight tooltip (and any other long tooltip) rendered as a single ~130-character line off the edge of the viewport (#498).
The pinned-tag insight badge no longer clips in the containers list. Centered with no max-width, it could overflow its narrow table column and get hard-clipped on both sides; it now truncates gracefully with a trailing ellipsis instead (#498).
Shortened the pinned-tag insight tooltip copy so it reads cleanly at the tooltip's new bounded width (#498).
Orphaned replacement container after a failed post-create network connect. createContainer created the replacement container and then connected it to any additional networks in the same try; if a network connect failed (for example a static-IP endpoint config rejected by a socket proxy), the error was rethrown without exposing the created container, so nothing could clean it up. The renamed original (<name>-old-<timestamp>) stayed parked while the orphaned Created-state replacement squatted the real name. Every rollback consumer — the regular Docker update executor, self-update, health-monitor auto-rollback, the backup-restore API, and Docker Compose rollback restore — now recovers the created container from the error and best-effort stops + force-removes it before restoring the original container's name, tolerating cleanup failures with a warning rather than masking the original error. The same recovery applies when the replacement is created successfully but then fails to start.
Repeated failed updates could cascade into a second rename. If a container was already left renamed -old-<timestamp> by a prior failed update, a subsequent update attempt against it now fails immediately with a clear "needs manual cleanup" error naming the true canonical name (even through nested renames) instead of renaming and recreating again on top of the unresolved failure. The same guard protects drydock's own self-update.
Replacement containers no longer inherit the previous container's auto-assigned MAC address. The recreate path copied each network endpoint's operational MacAddress straight from docker inspect, permanently re-pinning a MAC the daemon had generated (stale once the endpoint's IP changes, and rejected outright by MAC-denying socket proxies such as sockguard's default policy). A MAC is now carried over only when it was configured container-wide (docker run --mac-address / compose's service-level mac_address:, i.e. Config.MacAddress), and only onto the container's primary network; daemon-assigned MACs are left for the daemon to regenerate.
Locally-built images no longer spam nightly error counts. Containers running an image with no registry-hosted digest (built locally or docker loaded) were still queried against the registry on every watch cycle, producing a 401 that got counted as a watch error. Drydock now detects images with no RepoDigests at discovery/refresh time and skips the registry lookup for them entirely.
The digest-watch throttling warning no longer prints with domain undefined for unprefixed Docker Hub image references (e.g. nginx); it now shows docker.io.

Known limitations

If the rollback itself fails (for example the backup image can't be pulled, or the replacement container never becomes healthy), that failure is recorded in the update-operations store and audit log but does not yet trigger a push notification — check the container's operation history or the audit log after a failed update to confirm whether the rollback actually succeeded.
A per-network mac_address (compose networks.<net>.mac_address) set on a non-primary network cannot be distinguished from a daemon-assigned MAC via the Docker API today — the daemon persists the desired MAC internally but never exposes it in docker inspect output — so it is not preserved across recreates; the daemon assigns a fresh MAC for that network instead.

1.5.2-rc.5 — 2026-07-13

Fixed

Orphaned replacement container after a failed post-create network connect. createContainer created the replacement container and then connected it to any additional networks in the same try; if a network connect failed (for example a static-IP endpoint config rejected by a socket proxy), the error was rethrown without exposing the created container, so nothing could clean it up. The renamed original (<name>-old-<timestamp>) stayed parked while the orphaned Created-state replacement squatted the real name. Every rollback consumer — the regular Docker update executor, self-update, health-monitor auto-rollback, the backup-restore API, and Docker Compose rollback restore — now recovers the created container from the error and best-effort stops + force-removes it before restoring the original container's name, tolerating cleanup failures with a warning rather than masking the original error. The same recovery applies when the replacement is created successfully but then fails to start.
Repeated failed updates could cascade into a second rename. If a container was already left renamed -old-<timestamp> by a prior failed update, a subsequent update attempt against it now fails immediately with a clear "needs manual cleanup" error naming the true canonical name (even through nested renames) instead of renaming and recreating again on top of the unresolved failure. The same guard protects drydock's own self-update.
Replacement containers no longer inherit the previous container's auto-assigned MAC address. The recreate path copied each network endpoint's operational MacAddress straight from docker inspect, permanently re-pinning a MAC the daemon had generated (stale once the endpoint's IP changes, and rejected outright by MAC-denying socket proxies such as sockguard's default policy). A MAC is now carried over only when it was configured container-wide (docker run --mac-address / compose's service-level mac_address:, i.e. Config.MacAddress), and only onto the container's primary network; daemon-assigned MACs are left for the daemon to regenerate.
Locally-built images no longer spam nightly error counts. Containers running an image with no registry-hosted digest (built locally or docker loaded) were still queried against the registry on every watch cycle, producing a 401 that got counted as a watch error. Drydock now detects images with no RepoDigests at discovery/refresh time and skips the registry lookup for them entirely.
The digest-watch throttling warning no longer prints with domain undefined for unprefixed Docker Hub image references (e.g. nginx); it now shows docker.io.

Changed

Docs: documented the socket-proxy requirements for recreating containers with a static IP or MAC address (macvlan, custom networks). Tecnativa/docker-socket-proxy needs both POST=1 and NETWORKS=1 for containers on more than one network (the extra POST /networks/{id}/connect calls); sockguard needs request_body.network.allow_endpoint_config: true to permit endpoint configs that specify a static IP or MAC address — and as of sockguard v1.5.0 and later (forthcoming at the time of this release), that requirement also applies to POST /containers/create, so single-network macvlan/static-IP containers need the policy set too, not just multi-network ones (older sockguard versions enforce the policy only at network-connect time). Added a matching FAQ entry for containers left renamed -old-<timestamp> after a failed update, including a note for operators who intentionally name a container in a way that collides with drydock's own rollback naming convention.

Known limitations

If the rollback itself fails (for example the backup image can't be pulled, or the replacement container never becomes healthy), that failure is recorded in the update-operations store and audit log but does not yet trigger a push notification — check the container's operation history or the audit log after a failed update to confirm whether the rollback actually succeeded.
A per-network mac_address (compose networks.<net>.mac_address) set on a non-primary network cannot be distinguished from a daemon-assigned MAC via the Docker API today — the daemon persists the desired MAC internally but never exposes it in docker inspect output — so it is not preserved across recreates; the daemon assigns a fresh MAC for that network instead.

1.5.2-rc.4 — 2026-07-12

Fixed

Tooltip text now wraps inside a bounded popup instead of rendering one screen-wide line — the shared tooltip directive had white-space: nowrap and no max-width, so the pinned-tag "Newer available" insight tooltip (and any other long tooltip) rendered as a single ~130-character line off the edge of the viewport (#498).
The pinned-tag insight badge no longer clips in the containers list. Centered with no max-width, it could overflow its narrow table column and get hard-clipped on both sides; it now truncates gracefully with a trailing ellipsis instead (#498).
Shortened the pinned-tag insight tooltip copy so it reads cleanly at the tooltip's new bounded width (#498).

1.5.2-rc.3 — 2026-07-11

Added

Informational version visibility for pinned tags (#498). A specific-precision, unlabeled, non-loose pinned tag (e.g. nginx:1.25.3) still gets digest-only comparison for update actions — that pin-gate behavior from rc.2 is unchanged — but the container result now carries a new updateInsight: { tag, kind } field showing the best newer same-family tag that exists in the registry, purely as information. It reuses the exact same strict-style family matching used for actionable updates (prefix + suffix/variant compatibility + matching numeric-segment count + CalVer leading-zero rules) — no exception is carved out for major-version jumps; strict matching never restricted those to begin with. One narrow widening is scoped to this informational channel only: a prerelease-pinned tag (e.g. 1.5.2-rc.1) can see its own bare GA release (1.5.2) here, but that never makes the bare GA release an actionable update candidate — the actionable path rejects it just as before, even under dd.tag.family=loose or a permissive dd.tag.include filter. What's new is exposing this comparison at all for pinned tags, since the pin gate itself already blocks acting on any of it. Ties at the same numeric version prefer the tag whose suffix template exactly matches the pinned tag's. This is additive only: updateAvailable, updateKind, and trigger dispatch are all unaffected, so nothing new fires because of it. On by default; opt out per watcher with DD_WATCHER_{name}_TAG_PIN_INFO=false. Surfaced in the UI as an informational "Newer available" badge next to the existing update-state badges.

Fixed

dd.tag.family=loose no longer bypasses the suffix/variant guard in isSemverFamilyMatch(). A pinned nginx:1.2.3-ls132 container could previously be offered a bare 1.2.4 (wrong variant) as an update candidate under loose policy — loose mode was only ever meant to relax prefix equality and CalVer leading-zero rules, not let updates cross a suffix/variant boundary entirely. The guard now applies unconditionally regardless of policy.
The candidate sort in sortSemverDescending() now prefers the exact-suffix-template match over a merely-compatible one when two candidates tie at the same numeric version (e.g. preferring 1.2.5-alpine over 1.2.5-alpine3.21 for a 1.2.3-alpine reference). Semver treats the suffix as a prerelease field, so without this fix the wrong variant could outrank the exact match purely on prerelease-string ordering.

1.5.2-rc.2 — 2026-07-10

Fixed

Pinned semver tags (e.g. nginx:1.25.3) are now compared by digest by default, as originally announced in v1.5.0-rc.36 — a rebuilt image republished under the same tag is detected again. Previously digest watching was silently disabled for fully-pinned tags, leaving them with no update detection at all and a misleading "compared by digest only" notice on every pinned container (#498). Version climbing for pinned tags remains opt-in via dd.tag.include or dd.tag.family=loose. Note: in agent deployments, agents perform the registry checks — update agents alongside the controller to restore digest detection for agent-watched containers.
The "no update detection" notice shown when dd.watch.digest=false is explicitly set no longer claims digest comparison is happening.
Removed the unreachable flat tagFamily imgset config key (env-derived keys are lowercased, so it could never match); the documented DD_WATCHER_{watcher}_IMGSET_{name}_TAG_FAMILY form is unaffected.

1.5.2-rc.1 — 2026-07-10

Fixed

Container update policy is no longer lost when a container is recreated (#496). Container documents are keyed by Docker's container ID, which changes every time a container is recreated (image pull, docker compose up -d, or an update trigger firing). The replacement was stored as a brand-new document and the per-container update policy — maturity gate, skipped tags/digests, snooze — was silently dropped along with the old one. Because an absent policy means "no gating" rather than "default gating", affected containers then updated immediately instead of respecting their maturity soak. The policy now survives a recreate, for containers watched locally and through a remote agent alike. Present since 1.4.1, when per-container maturity policies were introduced.
The remote-agent prune path now distinguishes a recreated container from a removed one, so a container that reappears under a new ID keeps its update policy and its Home Assistant state topic, while a genuinely deleted container still has its discovery topics cleaned up.

1.5.1 — 2026-07-09

Consolidates the 1.5.1-rc.1 … 1.5.1-rc.6 prereleases. Users upgrading from 1.5.0 get everything below; users already on a 1.5.1 release candidate get the rc.6 fixes.

Added

Container software version in the detail panels and a new Version column in the containers table. Drydock now surfaces the application version baked into an image — read from the org.opencontainers.image.version OCI label, falling back to the running container's inspect metadata — as image.softwareVersion. It appears in the container side panel, the full-page detail view, and a new Version column in the containers table. The existing Tag column (column key version, preserved so saved column preferences keep working) continues to show the image tag; the new Version column shows image.softwareVersion, falling back to the tag when no software version is available. dd.inspect.tag.path now dual-writes the extracted value into image.softwareVersion as well as overwriting the image tag, so the Version column is populated for inspect-path containers with no label change needed. The Version column is visible by default for new installs; existing users have it inserted into their saved column list automatically on first load after upgrading. (#209)
dd.inspect.tag.version-only opt-in label. When dd.inspect.tag.path is set, the extracted value normally overwrites the image tag (enabling update detection against the semver embedded in the running container). Setting dd.inspect.tag.version-only=true routes the extracted value to image.softwareVersion only, leaving the real image tag intact for update detection. This is useful when the inspect path carries a displayable application version that differs in format from the registry tag — the Version column shows it without disrupting how drydock matches updates. The default (tag overwrite) is unchanged when the label is absent. (#209)
Intermediate release notes between the running and target version. When a container is several versions behind, drydock now fetches the releases between the running tag (exclusive) and the update target (inclusive) and shows them in the release-notes popover. Best-effort and semver-only — date tags and rolling tags (latest, stable) fall back to the standard two-panel view. Cap the range with DD_RELEASE_NOTES_MAX_INTERMEDIATE (default 20; set to 0 to disable). When the range exceeds the cap, the popover shows a non-silent "N older releases not shown" notice. Supports the __FILE secret-file convention (DD_RELEASE_NOTES_MAX_INTERMEDIATE__FILE). (#453)
New GET /api/containers/{id}/intermediate-release-notes endpoint. Lazy-loads the intermediate release list on demand when the release-notes popover opens; not embedded in the container model or agent snapshot, so it adds no ongoing payload weight. Accepts from (required) and to (defaults to the container's pending update tag) query parameters. (#453)
Optional mount-prefix fallback for Docker Compose path matching. When a watched container's resolved compose file path differs from the trigger's configured compose file only by a mount prefix (common with Portainer and bind-mounted compose files), drydock can now match on the trailing <project-dir>/<file> tail instead of skipping the container. Off by default — enable it per trigger with DD_ACTION_DOCKERCOMPOSE_<name>_MOUNT_PREFIX_FALLBACK=true. It stays opt-in because tail matching cannot distinguish two stacks that share a project-directory name across environments (e.g. /prod/myapp vs /staging/myapp). (#365)
$currentReleaseNotes trigger template variable. Trigger templates (notification bodies, command arguments, and the like) can now reference $currentReleaseNotes to include the release notes for the container's currently running version, alongside the existing variable for the update target's notes. (#295)
Container uptime. The side panel and full-page detail view now show how long a container has been running (from the Docker State.StartedAt timestamp), and a new opt-in Uptime column can be enabled in the containers table via the column picker. The value updates live and falls back to an em-dash when the start time is unknown.
Warn log when a dd.source.repo container label shadows a trusted OCI image source label. Adding a dd.source.repo label to a running container when the image already carries a trusted org.opencontainers.image.source (or org.opencontainers.image.url) OCI label silently downgrades source resolution from trusted to untrusted, which drops the GHCR token fallback for release-notes lookups. Drydock now logs a warn-level line each watch cycle when it detects this conflict, naming both repos. (#452)
Remote agents now report their log level and watcher schedule. The agent handshake (dd:ack) includes logLevel and pollInterval (the watcher's cron), so the GET /api/v1/agents response and the Agents view populate these fields for connected agents instead of leaving them blank.

Changed

The entire UI is now translatable. The last hardcoded English strings (dashboard widgets, security view, detail panels, host-status labels, the log viewer's invalid-regex notice, and SSE update-failed fallbacks) were extracted into the vue-i18n catalogs, so every surface now resolves through the translation system. Combined with the newly opened community translation project on Crowdin, contributors can translate any part of the interface. The 16 community locales ship with this release, synced from Crowdin.
Maturity gate counts from the registry publish date when trustworthy. For Docker Hub and GHCR (including lscr.io), the gate now measures elapsed time from the real image push date (last_updated / updated_at) instead of from when drydock first detected the update. An image that has been public longer than maturityMinAgeDays clears the gate immediately on the first scan that finds it. All other registries expose only the OCI image build date, which is not a reliable push signal, so drydock falls back to its own first-detection timestamp (updateDetectedAt) for those. A trusted publish date is skipped if it fails to parse or is in the future (clock-skew protection).
DD_RELEASE_NOTES_GITHUB_TOKEN is now forwarded to release-notes lookups for repos resolved from a dd.source.repo container label or a persisted container.sourceRepo value. Previously these sources were always fetched anonymously. The GHCR token fallback stays restricted to trusted sources (OCI image labels and GHCR image paths) and is never sent to a container-label source. Because the dedicated token can be sent to a repo named by a container label, scope it narrowly: a classic PAT with public_repo scope only, or a fine-grained PAT with read-only Contents permission limited to public repositories and no write or account permissions. (#452)
Container validation now tolerates fields written by newer drydock versions. The store validator no longer rejects unknown keys, so a dd.json written by a newer release stays readable after a downgrade. Note: this protects downgrades from v1.5.1 onward — rolling back from v1.5.1 to v1.5.0 (which predates this change) still requires removing the new details.startedAt and image.softwareVersion fields from dd.json, since v1.5.0 rejects them.
Coverage reporting moved from Codecov to Qlty Cloud. Part of the org-wide consolidation onto Qlty (one vendor for code quality and coverage). CI now publishes the normalized app/ui lcov reports to Qlty Cloud via GitHub OIDC — no stored coverage token — replacing the Codecov upload and codecov.yml. The vitest 100% coverage thresholds in the app/ and ui/ test suites remain the enforced gate; the README coverage badge now points at Qlty.

Fixed

Maturity gate (maturityMode: 'mature') never triggered; the first-detection timestamp was never computed. The function that stamps updateDetectedAt returned undefined immediately whenever updateAvailable was false, which is always the case while maturity suppression is active, so the timestamp was never computed and the maturity clock never started. Containers blocked by the maturity gate remained permanently "maturing" and never became update-available.
Maturity clock reset on every container recreation. When a container was stopped and recreated (Portainer stack redeploy, docker compose down && up), its new Docker container ID caused drydock to treat it as a fresh container, resetting updateDetectedAt to zero. Containers that are frequently redeployed could never accumulate enough age to clear the gate. The clock is now keyed on a stable container identity and survives recreation as long as the same update remains pending, including the common case where a slow image pull means the replacement container isn't yet visible when the old one is pruned.
A changed update candidate now restarts the maturity soak. When a new image digest or tag is published while an earlier update is still inside the maturity window, the gate restarts the clock for the new candidate instead of letting it inherit the previous candidate's elapsed time. A freshly pushed image always soaks for the full maturityMinAgeDays rather than being treated as already mature. (Local watches previously kept the original detection time here; remote-agent containers already behaved this way.)
In-memory container cache key collision. The cache key joined watcher name and container name with _, so my_prod + nginx and my + prod_nginx produced the same key. A wrong cache hit could apply a stale security scan result or maturity timestamp from one container to a different container, most visibly after a recreation event. The separator is now ::.
Docker and Docker Compose actions can pull private GCR and Google Artifact Registry images again. getAuthPull() for the GCR and GAR providers returned the raw service-account email as the username and the private key as the password, which docker login rejects, so any action trigger targeting a private gcr.io or *-docker.pkg.dev image failed to authenticate and could not apply the update. It now returns the _json_key username with the service-account JSON as the password, the format Google's registry auth expects (and the same one the token-exchange path already used).
Quay tag pagination no longer breaks on standard Link headers. The next_page cursor was matched with a greedy pattern that swallowed the trailing >; rel="next" from an RFC 5988 Link header and corrupted the cursor, so repositories with more than one page of tags could silently stop paginating. The parser now reads next_page and last cursors correctly from both bare-URL and RFC 5988 header forms, and still URL-encodes them to block scope injection. The inherited TrueForge provider gets the same fix.
Auto-apply update triggers now honor the maintenance window on every detection path. A container update detected through certain code paths could be auto-applied outside the configured maintenance window because the window check was missing on those paths. The gate is now enforced uniformly, so updates only auto-apply inside the window regardless of how the update was detected. (#321)
Home Assistant-style PEP 440 nightly tags (and other lossy version formats) no longer masquerade as a stable "suggested pin". Tag suggestions for latest/untagged containers relied on semver.coerce() as a last-resort parser, which silently drops any suffix it doesn't understand — a PEP 440 dev/post release (2026.8.0.dev202607050315, 1.2.3.post1), an OS-variant suffix (3.11-bullseye), or a hyphenated CalVer date (2024-01-15) all coerced down to a bare major.minor.patch and got offered as "stable". Suggestions now reject any candidate tag that required this lossy coercion unless the raw tag is itself a bare numeric version (optionally v-prefixed, 1-3 dot-separated groups) that provably lost nothing. The containers table also now renders the suggested-tag hint through the existing SuggestedTagBadge component (labeled "Suggested" with a tooltip) instead of an unlabeled raw string next to the Digest/NEW badges. (#473)
"Click to copy" did nothing and logged a TypeError on deployments served over plain HTTP (the common self-hosted LAN setup), because the browser Clipboard API only exists in secure contexts. Copying now falls back to the legacy execCommand technique when the API is missing or rejects, covers the log viewer's Copy button too, and shows a "Copy failed" state instead of failing silently when no copy mechanism works at all. (#472)
An open tooltip whose text changed — like the copy button's "Copied" confirmation — stayed stuck on the old text until you moved the mouse away and back. Clicking to copy hid the tooltip outright, so the "Copied"/"Copy failed" state never appeared until a re-hover. Tooltips now update their text in place while open, reposition themselves for the new content, and reappear immediately if the pointer never left. Three call sites that stacked a second tooltip directly onto a copyable tag's root element (the containers table's digest-delta tooltip and two spots in the dashboard's recent-updates widget) now feed their text into the tag's own tooltip through a new idleTooltip prop instead of clobbering it. (#472)
The security view now shows release notes for the running image even when no update is pending. The detail panel, table, and card surfaces previously gated the release-notes link behind "an update is available," so a container with no pending update showed nothing even though its current release notes were known. The running-tag notes now appear whenever they exist, and a "View project" link (the source repository) was added alongside them, matching the containers view. (Discussion #295)
The dashboard "Recent Updates" widget now uses the shared release-notes and project-link components. It previously rendered a bare release-notes anchor with no project link and no structured current/available notes. It now renders the same icon links as every other surface, fed from the container's sourceRepo, releaseNotes, and currentReleaseNotes. (Discussion #295)
Completed i18n coverage for the last untranslated UI surfaces. A code-level audit found several strings that still rendered in English for non-English users; they now resolve through the translation catalog: the trigger status badge (active/inactive), the running/writes-compose yes/no preview values, the "container actions disabled by server configuration" tooltip, the update-maturity "Available for N days" tooltip (the translate function is now threaded through the container mapper, which previously left the existing catalog keys unused), the grouped "Update All" success toast (which appended a raw English in <group> — it now interpolates the group name through a translatable key), the security-view severity tooltips (CRITICAL/HIGH/MEDIUM/LOW), the backup operation unknown fallback label, and the search-bar hint footer connectors. (#329)
Grouped "Update All" buttons could scroll out of view at moderate desktop widths once the new Version column widened the containers table: columns were only ever laid out at their preferred widths, so the per-stack Update All button — positioned at the far end of the group header row — ended up past the visible edge while everything else looked normal. Tables now shrink columns proportionally toward their minimum widths when space is tight, and the group header's Update All button is additionally pinned to the visible edge, so it stays reachable even when a table legitimately overflows. (Affected the 1.5.1-rc.4/rc.5 prereleases only.) (#467)

Security

Base image refreshed to clear 24 container-scan CVEs. Bumped the pinned node:24-alpine base from a stale digest (Node 24.16.0, Alpine 3.21) to the current digest (Node 24.18.0, Alpine 3.24) and added libexpat to the targeted apk upgrade set. This resolves all 11 Node binary CVEs reported by the image scan — including the one critical (CVE-2026-48930) and four high — plus 13 medium libexpat CVEs (now 2.8.2-r0). A rebuild + rescan confirms zero critical/high/Node/libexpat findings remain. The three busybox/ssl_client findings (CVE-2025-60876, medium) have no upstream fix in Alpine yet and are tracked for a later base bump. All previously pinned Alpine package versions still resolve on 3.24, so the build is otherwise unchanged.
Custom registry TLS settings now apply to every registry request. DD_REGISTRY_*_CAFILE, _INSECURE, and _CLIENTCERT were honored only on the credential handshake for the GAR, GitLab, Mau, DHI, ACR, and ECR providers; the follow-up tag-list, manifest, and blob calls fell back to the system CA bundle. The custom TLS agent is now propagated to those calls, including the anonymous (no-credential) code paths in GAR, GCR, and Quay, so a private CA or an insecure setting is enforced end to end rather than only while fetching the token.
Hook command environment values are sanitized against shell injection. Lifecycle hook commands run through /bin/sh -c, and registry-controlled values (image name, tag, update digest) flowed into the hook environment unsanitized, so a crafted tag could inject shell commands into a hook script that expanded those variables unquoted. The values are now scrubbed of shell metacharacters, matching the sanitization the command action already applies.
DD_SESSION_SECRET__FILE is now honored. The session secret was read straight from process.env, bypassing the __FILE secret-file resolution every other secret uses, so the documented file form was silently ignored and the instance fell back to a generated secret on every restart. It now reads the resolved value, so a secret supplied via DD_SESSION_SECRET__FILE is used (and, as with every other secret, the file form wins when both the file and the bare variable are set).
Secret files are checked for unsafe permissions and trailing newlines. When a DD_*__FILE secret is readable by group or others, drydock logs a non-fatal warning recommending chmod 600 (skipped on Windows, where the mode bits are not meaningful). File-sourced secret values are also trimmed of a trailing newline so an editor- or echo-added \n can't corrupt a credential, matching the common Docker *_FILE convention.
Debug dump and container environment no longer leak credentials. The debug dump's redaction missed SMTP passwords (*_PASS keys) and webhook URLs with embedded secrets, both of which appeared in the dumped environment for any authenticated user. They are now redacted, and the same *_PASS gap is closed for the container runtime environment shown via /api/containers. Registry usernames and service hostnames stay visible by design, since they aren't secrets and aid debugging.
Suppressed a ZAP DAST false positive (rule 10049, Storable and Cacheable Content). The baseline scan flagged cacheable static responses that are not sensitive; the rule is now downgraded in .zap/rules.tsv and the JSON-to-SARIF converter handles the suppression so the security workflow stays green without masking real findings. (#374)

Upgrade Notes

One-time notification burst on first scan after upgrade (Docker Hub / GHCR containers only). Containers on Docker Hub or GHCR whose pending update is already older than maturityMinAgeDays will clear the maturity gate immediately on the first poll after upgrading. Notification triggers in always mode will fire once for each such container. Action triggers (docker, docker-compose, command) will also fire, so containers previously held by the gate may be updated automatically on that first poll. Review your active action-trigger configuration before upgrading if you want to control the timing. This is expected behavior: those images were already mature; the gate was simply unaware of it.

Known Issues

Home Assistant update.dd_container entities can show "Unknown" via the MQTT + HASS trigger. The latest_version template renders empty when a container has no pending update, blanking the entity instead of reporting that it is up to date. Pre-existing; the fix ships in v1.6. (#491)

1.5.1-rc.6 — 2026-07-05

Fixed

Home Assistant-style PEP 440 nightly tags (and other lossy version formats) no longer masquerade as a stable "suggested pin". Tag suggestions for latest/untagged containers relied on semver.coerce() as a last-resort parser, which silently drops any suffix it doesn't understand — a PEP 440 dev/post release (2026.8.0.dev202607050315, 1.2.3.post1), an OS-variant suffix (3.11-bullseye), or a hyphenated CalVer date (2024-01-15) all coerced down to a bare major.minor.patch and got offered as "stable". Suggestions now reject any candidate tag that required this lossy coercion unless the raw tag is itself a bare numeric version (optionally v-prefixed, 1-3 dot-separated groups) that provably lost nothing. The containers table also now renders the suggested-tag hint through the existing SuggestedTagBadge component (labeled "Suggested" with a tooltip) instead of an unlabeled raw string next to the Digest/NEW badges. (#473)
"Click to copy" did nothing and logged a TypeError on deployments served over plain HTTP (the common self-hosted LAN setup), because the browser Clipboard API only exists in secure contexts. Copying now falls back to the legacy execCommand technique when the API is missing or rejects, covers the log viewer's Copy button too, and shows a "Copy failed" state instead of failing silently when no copy mechanism works at all. (#472)
An open tooltip whose text changed — like the copy button's "Copied" confirmation — stayed stuck on the old text until you moved the mouse away and back. Clicking to copy hid the tooltip outright, so the "Copied"/"Copy failed" state never appeared until a re-hover. Tooltips now update their text in place while open, reposition themselves for the new content, and reappear immediately if the pointer never left. (#472)
Three call sites bound a second tooltip directly onto a copyable tag's root element (the containers table's digest-delta tooltip and two spots in the dashboard's recent-updates widget), so it silently clobbered the tag's own copy-state tooltip — duplicate listeners and, after the content-change re-show above, occasional spurious re-shows while hovering. The informative text now flows into the tag's own tooltip via a new idleTooltip prop instead of stacking a second directive on the same element. (#472)

1.5.1-rc.5 — 2026-07-02

Fixed

Grouped "Update All" buttons could scroll out of view after upgrading to rc.4. The new Version column added in rc.4 widened the containers table for existing users, and at moderate desktop widths the table overflowed horizontally in a way its responsive column-hiding never accounted for: columns were only ever laid out at their preferred widths, so the per-stack Update All button — positioned at the far end of the group header row — ended up past the visible edge while everything else looked normal. Tables now shrink columns proportionally toward their minimum widths when space is tight (matching the widths the responsive logic already budgets for), and the group header's Update All button is additionally pinned to the visible edge, so it stays reachable even when a table legitimately overflows. (#467)

1.5.1-rc.4 — 2026-06-29

Added

Optional mount-prefix fallback for Docker Compose path matching. When a watched container's resolved compose file path differs from the trigger's configured compose file only by a mount prefix (common with Portainer and bind-mounted compose files), drydock can now match on the trailing <project-dir>/<file> tail instead of skipping the container. Off by default — enable it per trigger with DD_ACTION_DOCKERCOMPOSE_<name>_MOUNTPREFIXFALLBACK=true. It stays opt-in because tail matching cannot distinguish two stacks that share a project-directory name across environments (e.g. /prod/myapp vs /staging/myapp). (#365)
${currentReleaseNotes} trigger template variable. Trigger templates (notification bodies, command arguments, and the like) can now reference ${currentReleaseNotes} to include the release notes for the container's currently running version, alongside the existing variable for the update target's notes. (#295)
Container software version in the detail panels and a new Version column in the containers table. Drydock now surfaces the application version baked into an image — read from the org.opencontainers.image.version OCI label, falling back to the running container's inspect metadata — as image.softwareVersion. It appears in the container side panel, the full-page detail view, and a new Version column in the containers table. The existing Tag column (column key version, preserved so saved column preferences keep working) continues to show the image tag; the new Version column shows image.softwareVersion, falling back to the tag when no software version is available. dd.inspect.tag.path now dual-writes the extracted value into image.softwareVersion as well as overwriting the image tag, so the Version column is populated for inspect-path containers with no label change needed. The Version column is visible by default for new installs; existing users have it inserted into their saved column list automatically on first load after upgrading. (#209)
dd.inspect.tag.version-only opt-in label. When dd.inspect.tag.path is set, the extracted value normally overwrites the image tag (enabling update detection against the semver embedded in the running container). Setting dd.inspect.tag.version-only=true routes the extracted value to image.softwareVersion only, leaving the real image tag intact for update detection. This is useful when the inspect path carries a displayable application version that differs in format from the registry tag — the Version column shows it without disrupting how drydock matches updates. The default (tag overwrite) is unchanged when the label is absent. (#209)
Container uptime. The side panel and full-page detail view now show how long a container has been running (from the Docker State.StartedAt timestamp), and a new opt-in Uptime column can be enabled in the containers table via the column picker. The value updates live and falls back to an em-dash when the start time is unknown.

Changed

Container validation now tolerates fields written by newer drydock versions. The store validator no longer rejects unknown keys, so a dd.json written by a newer release stays readable after a downgrade. Note: this protects downgrades from v1.5.1 onward — rolling back from v1.5.1 to v1.5.0 (which predates this change) still requires removing the new details.startedAt and image.softwareVersion fields from dd.json, since v1.5.0 rejects them.

Fixed

Completed i18n coverage for the last untranslated UI surfaces. A code-level audit found several strings that still rendered in English for non-English users; they now resolve through the translation catalog: the trigger status badge (active/inactive), the running/writes-compose yes/no preview values, the "container actions disabled by server configuration" tooltip, the update-maturity "Available for N days" tooltip (the translate function is now threaded through the container mapper, which previously left the existing catalog keys unused), the grouped "Update All" success toast (which appended a raw English in <group> — it now interpolates the group name through a translatable key), the security-view severity tooltips (CRITICAL/HIGH/MEDIUM/LOW), the backup operation unknown fallback label, and the search-bar hint footer connectors. The new English catalog keys ship now; the 16 community locales fill in through the normal Crowdin sync after release. (#329)

Security

Base image refreshed to clear 24 container-scan CVEs. Bumped the pinned node:24-alpine base from a stale digest (Node 24.16.0, Alpine 3.21) to the current digest (Node 24.18.0, Alpine 3.24) and added libexpat to the targeted apk upgrade set. This resolves all 11 Node binary CVEs reported by the image scan — including the one critical (CVE-2026-48930) and four high — plus 13 medium libexpat CVEs (now 2.8.2-r0). A rebuild + rescan confirms zero critical/high/Node/libexpat findings remain. The three busybox/ssl_client findings (CVE-2025-60876, medium) have no upstream fix in Alpine yet and are tracked for a later base bump. All previously pinned Alpine package versions still resolve on 3.24, so the build is otherwise unchanged.

1.5.1-rc.3 — 2026-06-28

Added

Intermediate release notes between the running and target version. When a container is several versions behind, drydock now fetches the releases between the running tag (exclusive) and the update target (inclusive) and shows them in the release-notes popover. Best-effort and semver-only — date tags and rolling tags (latest, stable) fall back to the standard two-panel view. Cap the range with DD_RELEASE_NOTES_MAX_INTERMEDIATE (default 20; set to 0 to disable). When the range exceeds the cap, the popover shows a non-silent "N older releases not shown" notice. Supports the __FILE secret-file convention (DD_RELEASE_NOTES_MAX_INTERMEDIATE__FILE). (#453)
New GET /api/containers/{id}/intermediate-release-notes endpoint. Lazy-loads the intermediate release list on demand when the release-notes popover opens; not embedded in the container model or agent snapshot, so it adds no ongoing payload weight. Accepts from (required) and to (defaults to the container's pending update tag) query parameters. (#453)
Warn log when a dd.source.repo container label shadows a trusted OCI image source label. Adding a dd.source.repo label to a running container when the image already carries a trusted org.opencontainers.image.source (or org.opencontainers.image.url) OCI label silently downgrades source resolution from trusted to untrusted, which drops the GHCR token fallback for release-notes lookups. Drydock now logs a warn-level line each watch cycle when it detects this conflict, naming both repos. (#452)

Changed

DD_RELEASE_NOTES_GITHUB_TOKEN is now forwarded to release-notes lookups for repos resolved from a dd.source.repo container label or a persisted container.sourceRepo value. Previously these sources were always fetched anonymously. The GHCR token fallback stays restricted to trusted sources (OCI image labels and GHCR image paths) and is never sent to a container-label source. Because the dedicated token can be sent to a repo named by a container label, scope it narrowly: a classic PAT with public_repo scope only, or a fine-grained PAT with read-only Contents permission limited to public repositories and no write or account permissions. (#452)
Re-synced the UI translation catalogs from Crowdin. The 16 target-locale containerComponents.json files were regenerated from the Crowdin project so their key order tracks the English source catalog, keeping the on-disk catalogs and the translation platform in lockstep as community translations land.

1.5.1-rc.2 — 2026-06-28

Changed

The entire UI is now translatable. The last hardcoded English strings (dashboard widgets, security view, detail panels, host-status labels, the log viewer's invalid-regex notice, and SSE update-failed fallbacks) were extracted into the vue-i18n catalogs, so every surface now resolves through the translation system. Combined with the newly opened community translation project on Crowdin, contributors can translate any part of the interface.

Fixed

The security view now shows release notes for the running image even when no update is pending. The detail panel, table, and card surfaces previously gated the release-notes link behind "an update is available," so a container with no pending update showed nothing even though its current release notes were known. The running-tag notes now appear whenever they exist, and a "View project" link (the source repository) was added alongside them, matching the containers view. (Discussion #295)
The dashboard "Recent Updates" widget now uses the shared release-notes and project-link components. It previously rendered a bare release-notes anchor with no project link and no structured current/available notes. It now renders the same icon links as every other surface, fed from the container's sourceRepo, releaseNotes, and currentReleaseNotes. (Discussion #295)
Auto-apply update triggers now honor the maintenance window on every detection path. A container update detected through certain code paths could be auto-applied outside the configured maintenance window because the window check was missing on those paths. The gate is now enforced uniformly, so updates only auto-apply inside the window regardless of how the update was detected. (#321)

Security

Suppressed a ZAP DAST false positive (rule 10049, Storable and Cacheable Content). The baseline scan flagged cacheable static responses that are not sensitive; the rule is now downgraded in .zap/rules.tsv and the JSON-to-SARIF converter handles the suppression so the security workflow stays green without masking real findings. (#374)

1.5.1-rc.1 — 2026-06-26

Added

Remote agents now report their log level and watcher schedule. The agent handshake (dd:ack) includes logLevel and pollInterval (the watcher's cron), so the GET /api/v1/agents response and the Agents view populate these fields for connected agents instead of leaving them blank.

Changed

Coverage reporting moved from Codecov to Qlty Cloud. Part of the org-wide consolidation onto Qlty (one vendor for code quality and coverage). CI now publishes the normalized app/ui lcov reports to Qlty Cloud via GitHub OIDC — no stored coverage token — replacing the Codecov upload and codecov.yml. The vitest 100% coverage thresholds in the app/ and ui/ test suites remain the enforced gate; the README coverage badge now points at Qlty.
Maturity gate counts from the registry publish date when trustworthy. For Docker Hub and GHCR (including lscr.io), the gate now measures elapsed time from the real image push date (last_updated / updated_at) instead of from when drydock first detected the update. An image that has been public longer than maturityMinAgeDays clears the gate immediately on the first scan that finds it. All other registries expose only the OCI image build date, which is not a reliable push signal, so drydock falls back to its own first-detection timestamp (updateDetectedAt) for those. A trusted publish date is skipped if it fails to parse or is in the future (clock-skew protection).

Fixed

Maturity gate (maturityMode: 'mature') never triggered; the first-detection timestamp was never computed. The function that stamps updateDetectedAt returned undefined immediately whenever updateAvailable was false, which is always the case while maturity suppression is active, so the timestamp was never computed and the maturity clock never started. Containers blocked by the maturity gate remained permanently "maturing" and never became update-available.
Maturity clock reset on every container recreation. When a container was stopped and recreated (Portainer stack redeploy, docker compose down && up), its new Docker container ID caused drydock to treat it as a fresh container, resetting updateDetectedAt to zero. Containers that are frequently redeployed could never accumulate enough age to clear the gate. The clock is now keyed on a stable container identity and survives recreation as long as the same update remains pending, including the common case where a slow image pull means the replacement container isn't yet visible when the old one is pruned.
In-memory container cache key collision. The cache key joined watcher name and container name with _, so my_prod + nginx and my + prod_nginx produced the same key. A wrong cache hit could apply a stale security scan result or maturity timestamp from one container to a different container, most visibly after a recreation event. The separator is now ::.
A changed update candidate now restarts the maturity soak. When a new image digest or tag is published while an earlier update is still inside the maturity window, the gate restarts the clock for the new candidate instead of letting it inherit the previous candidate's elapsed time. A freshly pushed image always soaks for the full maturityMinAgeDays rather than being treated as already mature. (Local watches previously kept the original detection time here; remote-agent containers already behaved this way.)
Docker and Docker Compose actions can pull private GCR and Google Artifact Registry images again. getAuthPull() for the GCR and GAR providers returned the raw service-account email as the username and the private key as the password, which docker login rejects, so any action trigger targeting a private gcr.io or *-docker.pkg.dev image failed to authenticate and could not apply the update. It now returns the _json_key username with the service-account JSON as the password, the format Google's registry auth expects (and the same one the token-exchange path already used).
Quay tag pagination no longer breaks on standard Link headers. The next_page cursor was matched with a greedy pattern that swallowed the trailing >; rel="next" from an RFC 5988 Link header and corrupted the cursor, so repositories with more than one page of tags could silently stop paginating. The parser now reads next_page and last cursors correctly from both bare-URL and RFC 5988 header forms, and still URL-encodes them to block scope injection. The inherited TrueForge provider gets the same fix.

Security

Custom registry TLS settings now apply to every registry request. DD_REGISTRY_*_CAFILE, _INSECURE, and _CLIENTCERT were honored only on the credential handshake for the GAR, GitLab, Mau, DHI, ACR, and ECR providers; the follow-up tag-list, manifest, and blob calls fell back to the system CA bundle. The custom TLS agent is now propagated to those calls, including the anonymous (no-credential) code paths in GAR, GCR, and Quay, so a private CA or an insecure setting is enforced end to end rather than only while fetching the token.
Hook command environment values are sanitized against shell injection. Lifecycle hook commands run through /bin/sh -c, and registry-controlled values (image name, tag, update digest) flowed into the hook environment unsanitized, so a crafted tag could inject shell commands into a hook script that expanded those variables unquoted. The values are now scrubbed of shell metacharacters, matching the sanitization the command action already applies.
DD_SESSION_SECRET__FILE is now honored. The session secret was read straight from process.env, bypassing the __FILE secret-file resolution every other secret uses, so the documented file form was silently ignored and the instance fell back to a generated secret on every restart. It now reads the resolved value, so a secret supplied via DD_SESSION_SECRET__FILE is used (and, as with every other secret, the file form wins when both the file and the bare variable are set).
Secret files are checked for unsafe permissions and trailing newlines. When a DD_*__FILE secret is readable by group or others, drydock logs a non-fatal warning recommending chmod 600 (skipped on Windows, where the mode bits are not meaningful). File-sourced secret values are also trimmed of a trailing newline so an editor- or echo-added \n can't corrupt a credential, matching the common Docker *_FILE convention.
Debug dump and container environment no longer leak credentials. The debug dump's redaction missed SMTP passwords (*_PASS keys) and webhook URLs with embedded secrets, both of which appeared in the dumped environment for any authenticated user. They are now redacted, and the same *_PASS gap is closed for the container runtime environment shown via /api/containers. Registry usernames and service hostnames stay visible by design, since they aren't secrets and aid debugging.

Upgrade Notes

One-time notification burst on first scan after upgrade (Docker Hub / GHCR containers only). Containers on Docker Hub or GHCR whose pending update is already older than maturityMinAgeDays will clear the maturity gate immediately on the first poll after upgrading. Notification triggers in always mode will fire once for each such container. Action triggers (docker, docker-compose, command) will also fire, so containers previously held by the gate may be updated automatically on that first poll. Review your active action-trigger configuration before upgrading if you want to control the timing. This is expected behavior: those images were already mature; the gate was simply unaware of it.

1.5.0 — 2026-06-22

Added

Trigger taxonomy split — DD_ACTION_* and DD_NOTIFICATION_* prefixes. Action triggers (Docker, Docker Compose, Command) now use DD_ACTION_* / dd.action.*; messaging triggers (Slack, SMTP, Discord, Telegram, ntfy, Pushover, and all others) use DD_NOTIFICATION_* / dd.notification.*. All three families remain interchangeable at runtime through v1.7.0. A migration CLI (drydock config migrate --source trigger) rewrites existing configs automatically. The legacy DD_TRIGGER_* / dd.trigger.* aliases are deprecated (removal targeted v1.7.0).
Experimental Portwing edge-agent mode. Agents behind NAT or firewalls can dial out to the controller over a persistent wss:// WebSocket instead of requiring an inbound connection. Enable with DD_EXPERIMENTAL_PORTWING=true. Uses Ed25519 public-key challenge-response auth; operator key management is exposed at /api/v1/portwing/keys.
Real-time container log viewer. WebSocket-based live log streaming from Docker containers in the UI — ANSI color rendering, automatic JSON pretty-printing, free-text/regex search, stdout/stderr and log-level filtering, copy to clipboard, and gzip download. Available in the container detail panel and at /containers/:id/logs.
Notification outbox with retry and dead-letter queue. Failed notification deliveries are persisted and retried with exponential backoff + jitter. After 5 attempts (configurable), entries move to a dead-letter queue. New /api/notifications/outbox REST surface and a dedicated UI page let operators list, retry, or discard entries.
Durable update queue with restart recovery. Container updates are queued server-side with per-trigger concurrency limits. Queued and mid-pull operations survive controller restarts and are recovered automatically. New cancel endpoint (POST /api/operations/:id/cancel) accepts both queued and in-flight operations. A global concurrent-update cap (DD_UPDATE_MAX_CONCURRENT) is available.
Customizable dashboard. Drag-to-reorder, resize, and per-widget visibility toggles. New Resource Usage widget shows fleet-wide CPU and memory with top-N consumers, fed by a live fleet-stats SSE stream (GET /api/v1/stats/summary/stream).
Diagnostic debug dump. One-click export of redacted system state from Configuration > Diagnostics — runtime metadata, component state, Docker diagnostics, recent events, and DD_* env vars, with sensitive values auto-redacted. Available at GET /api/v1/debug/dump.
SSE Last-Event-ID replay. Every broadcast event carries a monotonic <bootId>:<counter> id; clients reconnecting with Last-Event-ID receive missed events from a 5-minute ring buffer. Clients that fall behind the buffer receive a dd:resync-required event.
Update-eligibility blockers on container rows. Sixteen structured blocker reasons are surfaced inline on the Containers list, so users see why a container isn't updating without opening the detail drawer. Hard blockers lock the Update button; soft blockers show a warn-and-confirm modal.
Per-agent Home Assistant MQTT topic segmentation (DD_NOTIFICATION_MQTT_<name>_HASS_AGENTTOPICSEGMENT). Prevents two agents sharing the default watcher name from overwriting each other's topics. Opt-in for v1.5.x; targets default-on in v1.7.0.
Full i18n coverage across the UI. All hardcoded strings migrated to per-namespace JSON catalogs with 17 locale options (de, es, fr, it, nl, pl, pt-BR, tr, zh-CN, zh-TW, ar, ja, ko, ru, uk, vi, plus English). Human translations synced from Crowdin.
Colored startup banner. Renders the whale logo as a truecolor half-block art print on interactive terminals; auto-suppressed when not a TTY or when NO_COLOR is set.

Changed

Action trigger default mode changed to AUTO=oninclude. Action triggers no longer auto-update every container by default; an explicit dd.action.include label is required. Notification triggers are unaffected.
Default watcher cron relaxed from hourly to every 6 hours. Reduces registry pressure for the common case; override with DD_WATCHER_{name}_CRON.
Consolidated security scanning on Grype; dropped Snyk. Grype now scans both the built container image and all six npm lockfiles. Snyk's GitHub SCM integration, .snyk policy file, security-snyk-weekly.yml, and related scripts are removed. Free gates (CodeQL, dependency-review, OpenSSF Scorecard, zizmor) continue to run on every PR.
Healthcheck binary replaces curl. Default HEALTHCHECK now uses a 65 KB static C binary (/bin/healthcheck). curl is retained for user-defined overrides through v1.6.0 (removal in v1.7.0).
DD_SESSION_SECRET auto-generated and persisted when unset. On first boot without the variable set, drydock generates 64 random bytes and persists them in the store so sessions survive restarts. The env var still takes precedence.
Tag-last release pipeline. The git tag is pushed only after the Docker image is built, pushed, signed (cosign), and attested (SLSA). If the tag exists, the image exists.

Fixed

Containers pinned to a fully-specified semver tag no longer climb to newer versions by default. Tags classified as specific precision now track digest changes only. Opt in to semver climbing with dd.tag.include or dd.tag.family=loose.
Multi-agent deployments no longer produce spurious 409 conflicts, cross-agent container contamination, or persistent "0 running containers" in the controller UI. Root causes: the controller's local watcher was pruning remote-agent rows; agent watcher snapshots were lost when the SSE stream was half-open mid-reconnect. Fixed by scoping store reads and key derivation to { agent, watcher } at every call site, and having agents replay the latest snapshot to each new SSE client immediately after dd:ack.
Self-update overlay holds until the swap is complete. Three compounding bugs (premature overlay dismissal, unreachable finalize callback, helper container appearing in the watcher list) are fixed. A new unauthenticated /api/v1/self-update/{operationId}/status endpoint lets the UI poll during the restart window.
Spurious "update failed" and "update available" notifications eliminated. False notifications arising from concurrent duplicate requests (409 race), post-update stale watcher scans, timed-out operation TTLs, and digest/batch path suppress-lifecycle gaps are all closed.
Live log viewer and WebSocket upgrades now work behind TLS-terminating reverse proxies. WebSocket upgrades and the log-stream origin check now honor X-Forwarded-Host/X-Forwarded-Proto when DD_SERVER_TRUSTPROXY is set.
Docker Compose update no longer destroys the running container on failure. A pre-flight architecture-compatibility check runs before removing the old container; if the recreate still fails, the original container is restored from its captured spec.
Registry 429/503 handling with Retry-After and per-host token bucket. Every registry HTTP call retries up to 3 times with exponential backoff honoring upstream Retry-After; a per-host token bucket prevents self-inflicted rate limiting during large cron cycles.
Standard Bearer token exchange now works for Chainguard, Codeberg/Forgejo, and other v2 registries that issue a WWW-Authenticate challenge. callRegistry implements the spec-compliant challenge-response flow with realm-host validation.

Security

Command trigger no longer inherits the full process environment. Child processes receive a fixed allowlist of system variables plus drydock-provided container vars. Additional variables can be whitelisted with DD_ACTION_COMMAND_{name}_ENV.
HTTP trigger blocks cloud metadata endpoints. Requests resolving to 169.254.0.0/16, fe80::/10, and related link-local ranges (including IPv4-mapped and IPv4-compatible spellings) are rejected before sending. Opt-out available via DD_NOTIFICATION_HTTP_{name}_ALLOWMETADATA=true.
CSRF same-origin enforcement hardened. Forwarded-host headers are now trusted only when Express trust proxy is enabled; /auth mutations (logout, remember) are now covered by the same-origin check; and DD_SERVER_TRUSTPROXY=true (all hops) now emits a startup warning recommending a hop count.
Security digest templates no longer evaluated as JavaScript. SECURITYDIGESTTITLE/SECURITYDIGESTBODY previously passed through new Function() — arbitrary code execution. The renderer now uses the same sandboxed ${…} interpolation engine as all other trigger templates.
Bearer token-endpoint requests now set maxRedirects: 0 to prevent credential exfiltration if a token endpoint returns a redirect. Applied to BaseRegistry and all providers that build their own credentialed token-fetch requests.
Container image CVE surface cleared. Bumped node:24-alpine base (24.14.0 → 24.16.0), cosign (2.6.3 → 3.0.6), musl, curl, and git, clearing all HIGH/CRITICAL findings in drydock-controlled packages. Residual HIGHs inside vendored Go binaries (cosign, trivy) are scoped out of the fail-on-HIGH gate via .grype.yaml.
LokiJS query-filter values constrained to primitives. ?key[$regex]=…-style operator injection (ReDoS via native RegExp, $ne/$gt operator injection) is neutralized at the store choke point for every caller.
Multiple CVE patches: undici bumped to 8.5.0 (8 advisories including CVE-2026-9675), nodemailer bumped to 9.0.1 (GHSA-p6gq-j5cr-w38f, CVSS 7.1), protobufjs to 7.6.3 (GHSA-xq3m-2v4x-88gg, critical), vite to 8.0.16, axios to 1.16.1 (CVE-2025-62718), and additional transitive patches across all six lockfiles.

1.5.0-rc.38 — 2026-06-19

Added

Colored startup banner. When drydock starts on an interactive terminal it now renders the whale logo as a compact truecolor half-block banner followed by a drydock v<version> · <mode> identity line. The art is baked from the master logo (drydock.png) at build time by scripts/gen-banner.mjs, so startup decodes no image. The banner is written to stderr and suppressed automatically when stdout/stderr is not a TTY or NO_COLOR is set, so logs and piped output stay clean.

Changed

Consolidated dependency/CVE scanning on Grype; dropped Snyk. Snyk's GitHub SCM integration scans the full dependency requirement graph across every package.json/package-lock.json in the repo rather than the resolved, shipped dependency set, so it over-reports advisories in transitive packages the lockfile never actually resolves to — noise on top of a redundant paid integration. Grype replaces it on both axes: it scans the built container image (the image's package catalog is the dependency set actually shipped) and the six npm lockfiles (root, app, ui, e2e, apps/demo, apps/web), matching the lockfile-resolved versions instead of the manifest graph, so it does not emit the requirement-graph false positives. The free gates already in CI cover the rest — CodeQL (SAST), dependency-review (new-dependency CVEs on PRs), OpenSSF Scorecard, and zizmor — so nothing else was needed (Trivy intentionally not added; drydock is TypeScript/Node, so the Go call-graph scanner govulncheck used on sibling repos does not apply here). The new security-grype.yml runs the dependency scan on pull requests (path-filtered to dependency/Dockerfile/workflow changes) plus a weekly cron and manual dispatch, builds and scans the container image on scheduled/manual runs, and uploads distinct-category SARIF to the GitHub Security tab. Removed the .snyk policy file, the security-snyk-weekly.yml workflow, the setup-snyk composite action, and the scripts/snyk-* gate/quota scripts.
Refreshed the drydock whale logo across the app, website, demo, and docs. A new master render replaces the brand mark everywhere — the in-app logo and favicons, the website/demo favicons, PWA icons, and OpenGraph cards, and the README/docs logos (including the dark-mode variant). All brand assets are now regenerated from a single master (drydock.png) via scripts/regenerate-brand-assets.sh. Filenames are unchanged, so the Home Assistant entity_picture URL contract is preserved.

Security

Documentation site (apps/web) js-yaml pinned to 4.2.0 (GHSA-h67p-54hq-rp68). fumadocs-mdx pulled js-yaml 4.1.1 transitively; an override forces the patched 4.2.0. Build-time dependency of the website only — not part of the shipped drydock image.
E2E load-test harness @opentelemetry/core pinned to 2.8.0 (CVE-2026-54285). artillery pulled @opentelemetry/core 2.7.1 transitively, vulnerable to unbounded memory allocation in W3C Baggage propagation; an override forces the patched 2.8.0. Test-only dependency — not part of the shipped drydock image.
Patched the container image's HIGH/CRITICAL CVE surface and scoped the Grype image gate. The first grype-image scan on main flagged a pre-existing CVE backlog that nothing had been scanning (Snyk Container never ran — no token was configured). Bumped the node:24-alpine base (node 24.14.0 → 24.16.0 clearing CVE-2026-21710, musl 1.2.5 → 1.2.6, curl 8.19.0 → 8.20.0, git 2.52.0 → 2.54.0) and cosign 2.6.3 → 3.0.6, which clears every HIGH/CRITICAL in the Node runtime and Alpine OS packages. The only residual HIGH/CRITICAL findings live inside the vendored Go module graphs compiled into the bundled cosign and trivy CLI binaries (drydock shells out to them for signature verification and container scanning) — those clear only when Alpine rebuilds the packages, so a documented .grype.yaml scopes the fail-on-HIGH image gate to the dependencies drydock controls (Node, OS packages, the app npm graph) and excludes the two tool-binary locations. cosign 3.0.6 keeps the verify --output json/--certificate-identity/--certificate-oidc-issuer/--key flags drydock's signature path uses.
Patched a batch of newly-disclosed undici CVEs across the runtime and tooling workspaces. osv-scanner flagged eight undici advisories disclosed in 2026 — CVE-2026-6733, CVE-2026-6734, CVE-2026-9675, CVE-2026-9678, CVE-2026-9679, CVE-2026-9697, CVE-2026-11525, and CVE-2026-12151. The shipped backend (app) carries undici as a direct dependency and was on 8.3.0, vulnerable to all eight — bumped to 8.5.0, the only release clearing the full set (CVE-2026-9675 is fixed solely in 8.5.0), and pinned in overrides as well. The dashboard build (ui) and the e2e load-test harness pulled undici 7.25.0/7.26.0 transitively; an overrides entry forces 7.28.0 (the patched 7.x line) in each — build- and test-only, not part of the shipped image.
Patched nodemailer to 9.0.1 (GHSA-p6gq-j5cr-w38f, CVSS 7.1). A message-level raw option bypassed nodemailer's disableFileAccess/disableUrlAccess guards, allowing arbitrary file read and full-response SSRF in the delivered message. drydock's SMTP trigger only calls createTransport/sendMail with plain from/to/subject/text fields and never passes raw, so the sink isn't reachable here — but the advisory affects every release through 9.0.0 with the fix landing only in 9.0.1, so the direct dependency in app is bumped from 8.0.10. The 8→9 major jump doesn't touch the stable createTransport/sendMail core drydock relies on.

1.5.0-rc.37 — 2026-06-15

Security

Patched a batch of newly-disclosed transitive CVEs across every workspace. osv-scanner flagged advisories disclosed 2026-06-15 in build- and test-time dependencies: vite (CVE-2026-53571, CVE-2026-53632), @babel/core (CVE-2026-49356), form-data (CVE-2026-12143), protobufjs (CVE-2026-54269), and ws (CVE-2026-48779). Each is pinned to a fixed version via an override (or a direct bump where the dependency is direct). js-yaml@3.14.2, reachable only through artillery's test-only load-test harness, is triaged as unreachable: its sole fix removes the safeLoad() API artillery still calls, and it parses only trusted in-repo configs.

Changed

Registry rate-limiter burst raised from 5 to 10 for ghcr.io and Docker Hub. The conservative burst allowance was tripping the limiter during legitimate request spikes (enumerating tags across many containers at once); the sustained rate (2 req/s) is unchanged.
Hardened the E2E/CI suite against transient flakes. Crash-prone real-application e2e fixtures (Home Assistant, Radarr) now run a keep-alive entrypoint so the watcher consistently discovers the full container set instead of intermittently seeing one short; the test-bootstrap readiness count is now exact and strict; and the Playwright container-detail helpers wait on real conditions rather than fixed timeouts. No shipped runtime behavior changes from this item.

1.5.0-rc.36 — 2026-06-15

Added

Experimental Portwing edge-agent mode — agents behind NAT or firewalls can now dial OUT to drydock over a persistent wss:// WebSocket instead of waiting for an inbound controller connection (PR #429, M5). The feature is experimental and opt-in: set DD_EXPERIMENTAL_PORTWING=true to enable it. When disabled, the endpoint is not mounted and the feature has zero runtime footprint. Once enabled, agents connect to WS /api/v1/portwing/ws using the portwing/1.0 subprotocol. Authentication is Ed25519 public-key challenge-response with timestamp + nonce replay protection (±60 s clock-skew window, 16 MB maximum frame size). Operator key management is exposed through a REST registry at /api/v1/portwing/keys (list, register, revoke). Because the feature is experimental the protocol and API surface may change in a future release without a deprecation notice.
Remote-agent runtime info now carries logLevel and pollInterval in the acknowledgement payload (PR #430, M4). Drydock threads these fields through buildRuntimeInfoFromAck and surfaces them in the Agents view alongside the existing runtime metadata.

Fixed

Containers pinned to a fully-specified semver tag (e.g. image:v1.13.3, 3+ numeric segments) no longer climb to newer versions by default (#321). Tags classified as specific precision are now treated as digest-only by default — getTagCandidates returns an empty tag list so updates track digest changes only, not semver version bumps. Opt in to semver climbing by setting dd.tag.include (restricts climbing to matching tags) or dd.tag.family=loose (unrestricted climbing as before). Floating tags (latest, 16-alpine, etc.) and 1–2-segment partial versions are unaffected.
Maintenance window now gates when auto-updates are applied, not just when update checks run (#321). Previously watchFromCron respected the window, but maybeFastResyncAfterUpdate (the post-update fast resync) called watchContainer unconditionally — allowing a triggered resync to detect a new image and dispatch an update outside the window. The fast resync now mirrors the same maintenance-window guard used by watchFromCron. Additionally, computeUpdateEligibility gains a maintenanceWindowOpen context field: when false, a soft maintenance-window-closed blocker is recorded in the eligibility result. Manual UI/API-triggered updates pass undefined and remain ungated.
Self-update overlay no longer flickers — the UI holds the "Applying Update" screen until the swap is actually complete. Three compounding defects made the self-update experience look broken: (1) the UI's connectivity probe treated any successful /auth/user response as "update finished", but the old server keeps answering during the image pull — so the overlay dismissed almost immediately, then the page died again when the old container actually stopped; (2) the in-progress self-update operation could never be recorded as completed: the finalize callback secret was regenerated per-process (the helper's POST always failed with 403 after the restart) and startup reconciliation expired the operation before the helper could finalize it; and (3) the transient drydock-self-update-<timestamp> helper container carried no watch-exclusion label, so it flashed into the container list with watch-by-default enabled. The finalize secret is now per-operation, with its SHA-256 hash persisted on the operation row so the restarted process can validate the helper's callback; startup reconciliation grants fresh in-progress self-update operations a 10-minute grace window (with expiry as the bounded fallback if the helper dies); a new unauthenticated GET /api/v1/self-update/{operationId}/status endpoint reports the operation state while the session is unavailable mid-restart; the UI polls that endpoint during a self-update and only reloads once the operation reaches a terminal state (succeeded, rolled-back, failed, or expired); and the helper container is labeled dd.watch=false so it never appears in the watcher's container list. The UI also closes its SSE connection when self-update mode begins (after the ack is delivered) — the browser's built-in EventSource retry would otherwise reconnect to the new server, clear the overlay before the swap was committed, and leave the SPA running stale pre-update assets. Dry-run mode no longer shows the overlay at all: the UI notification is skipped and the no-op operation is marked terminal immediately instead of lingering in-progress.

1.5.0-rc.35 — 2026-06-10

Fixed

False "update failed" notification when concurrent update requests race (#421). When a duplicate update request was rejected with HTTP 409 ("update already in progress") while the winning update was still in flight, the controller classified the conflict as a genuine failure — firing an "update failed" notification immediately followed by "updated successfully". The duplicate classifier now recognizes three benign signals instead of one: a recently-succeeded operation (as before), a 409 response whose body explicitly carries the active-update lock message ("Container update already queued/in progress" — authoritative even before the winner's state has propagated from a remote agent over SSE), and another active (queued or in-progress) operation for the same container and agent+watcher identity. The same reclassification now also covers the Docker-native rename path in ContainerUpdateExecutor, which previously marked the duplicate failed before the outer classifier could run.
Update operations could hang in-progress forever when deferred rollback reconciliation failed. The deferred reconciliation callback only logged a warning on error, leaving the operation permanently active and blocking all future updates for that container until restart. The operation is now terminalized as failed (self-update and already-terminal operations excluded, as elsewhere).
Spurious "update available" notifications around just-updated containers (#408 hardening). Three escape hatches in the post-update suppression mechanism are closed: suppression now keys on both the container ID and the watcher-scoped name, so the recreated container's new Docker ID can no longer dodge the check; the batch retry buffer consults suppression before re-queuing; and on startup the suppression set is re-seeded from update operations that succeeded within the last hour, so a controller restart between an update and the watcher's confirming scan no longer re-fires a stale notification. Suppression entries now also expire after one hour and are cleared on trigger deregistration, so containers deleted outright (or agents that never reconnect) can no longer leak entries for the life of the process.
Live log viewer returned 403 behind TLS-terminating reverse proxies even with DD_SERVER_TRUSTPROXY set. WebSocket upgrades bypass Express, so the log-stream origin check never honored trust-proxy. It now compares the browser origin against X-Forwarded-Host/X-Forwarded-Proto (first hop) when trust proxy is enabled — and remains byte-for-byte strict when it is not.
Containers of permanently removed agents lingered in the store forever. Startup now prunes container rows whose agent no longer matches any registered agent component. Rows of registered-but-currently-disconnected agents are untouched.
"Updated successfully" toast fired while the new container was still starting (#290 follow-up). The toast settled on the old container's removal event during a recreate; it now waits for the replacement container's arrival (replacementExpected removals are skipped, and the new container's ID rides the dd:update-applied payload as newContainerId), closing the status gap between the last update activity and the success notification. The toast dedup TTL also gained a 20% margin over the server's SSE replay buffer to prevent a boundary duplicate on reconnect.
Watcher enrichment failures that threw non-Error values leaked malformed entries into the container snapshot. Thrown non-Error values are now wrapped, counted as enrichment errors, and excluded.
GCR registry reported anonymous configurations as authenticated. Gcr.getAuthPull() now returns undefined without credentials, matching the other providers. GHCR 404 detection now checks the axios response status instead of matching error-message strings.

Security

Command trigger no longer inherits the full process environment. User-authored command scripts previously received every DD_* secret (registry tokens, notification tokens, agent secrets) via process.env. The child environment is now built from a fixed allowlist (PATH, HOME, SHELL, USER, LANG, LC_ALL, TZ, TMPDIR, TMP, TEMP) plus the drydock-provided container variables. Scripts that legitimately need more can name additional variables with the new DD_ACTION_COMMAND_{name}_ENV option (comma-separated).
Hook commands can be restricted to an allowlist of binaries (DD_HOOKS_ALLOWED_COMMANDS). With hooks enabled, dd.hook.pre/dd.hook.post labels could invoke any binary on the image. The new comma-separated allowlist matches the hook command's first token (basename, or exact path for entries containing /); when unset, behavior is unchanged and a one-time warning recommends configuring it.
HTTP trigger blocks cloud metadata endpoints. Requests resolving to link-local ranges (169.254.0.0/16 including 169.254.169.254, fe80::/10, fd00:ec2::254) are rejected before sending — including IPv4-mapped and IPv4-compatible IPv6 spellings of those ranges (::ffff:169.254.169.254, ::ffff:a9fe:a9fe, ::169.254.169.254), which would otherwise slip past the literal-IP check. Private-network and localhost targets remain fully supported — they are the normal self-hosted case. The rare legitimate link-local target can opt out via DD_NOTIFICATION_HTTP_{name}_ALLOWMETADATA=true.

Performance

Tag transform patterns are no longer recompiled in the sort hot path. Compiled RE2 transform patterns are cached per formula and tag candidates are transformed once before sorting instead of twice per comparison — previously ~3,000 compilations per 300-tag container per watch cycle.
Container normalization no longer deep-clones the entire container per registry call. The watcher now copies only the image/registry fields it mutates instead of structuredClone of labels and environment for every container every cycle.
Update-operation retention pruning uses indexed status queries instead of materializing the whole collection every 100th mutation, and the default rejected-credential pattern in BaseRegistry is compiled once at module load. An unused updatedAt collection index no longer taxes every mutation.

Changed

Trigger providers must implement trigger()/triggerBatch(). The base implementations now throw instead of silently doing nothing, so a provider that forgets to override fails loudly. All 22 bundled providers already comply; this only affects out-of-tree forks.

1.5.0-rc.34 — 2026-06-07

Added

Trigger environment variable taxonomy split — DD_ACTION_* and DD_NOTIFICATION_* prefixes. Action triggers (Docker, Docker Compose, Command) are now configured with DD_ACTION_* and dd.action.* labels; notification/messaging triggers (Slack, SMTP, Discord, Telegram, ntfy, Pushover, and all others) are configured with DD_NOTIFICATION_* and dd.notification.* labels. All three prefix families (DD_ACTION_*, DD_NOTIFICATION_*, DD_TRIGGER_*) are interchangeable at runtime — merge priority is DD_NOTIFICATION_* > DD_ACTION_* > DD_TRIGGER_*. A migration CLI (drydock config migrate --source trigger) rewrites DD_TRIGGER_*, dd.trigger.include, and dd.trigger.exclude to action-prefixed aliases automatically; use --dry-run to preview changes before applying.
Per-agent Home Assistant MQTT topic segmentation (DD_NOTIFICATION_MQTT_<name>_HASS_AGENTTOPICSEGMENT, default false). When enabled, Drydock inserts an agent/<name> segment into every Home Assistant MQTT topic — per-container state topics, watcher-level count/update sensors, and watcher running-status sensors — for containers owned by a remote agent, so two agents that both use the default watcher name local no longer publish to (and overwrite) the same topics. Enabling it also scopes the watcher-level sensor counts and the discovery-entity cleanup per agent, fixing the Home Assistant facet of #386. Controller-local container topics are unchanged. Because it changes the Home Assistant entity IDs for agent-owned containers, it is opt-in for the v1.5.x line and targeted to become the default in v1.7.0 — see Deprecated.
Up-to-date and pinned badges in Kind column — Containers table now shows a green check-circle badge ("Up to date") for containers at their latest version, and a green pin badge ("Pinned") for containers with skipped updates, replacing the previous dash placeholder.
Show/hide toggle on the login password field (commit e086c5bc). The sign-in password input now has an eye / eye-slash button to reveal or mask what was typed, with an accessible label and type="button" so it never submits the form.
Real-time container log viewer — WebSocket-based live log streaming from Docker containers directly in the UI. Features ANSI color rendering, automatic JSON log detection with syntax-highlighted pretty-printing, free-text and regex search with match navigation, stdout/stderr stream filtering, log level filtering for structured logs, copy to clipboard, and gzip-compressed download. Available in both the container detail panel and a dedicated full-page view at /containers/:id/logs. (Phase 4.2)
Diagnostic debug dump — One-click export of redacted system state from Configuration > Diagnostics. Collects runtime metadata, component state (watchers, registries, triggers, agents), Docker API diagnostics, MQTT Home Assistant sensors, recent Docker events, store stats, and DD_* environment variables. Sensitive values matching password|token|secret|key|hash are automatically redacted. Configurable time window (1–1440 minutes). (Phase 4.14)
Container log streaming API — WS /api/v1/containers/:id/logs/stream endpoint with Docker binary stream demultiplexing, session-based authentication on WebSocket upgrade, and fixed-window rate limiting (1,000 connections per 15 minutes).
Container log download API — GET /api/v1/containers/:id/logs endpoint with gzip compression support, stdout/stderr filtering, configurable tail size, and timestamp-based since filtering.
Debug dump API — GET /api/v1/debug/dump endpoint with configurable minutes query parameter for time-windowed event collection.
Dashboard customization — Customizable grid layout with drag-to-reorder, resize, and per-widget visibility toggles using grid-layout-plus. Edit mode via pencil icon in breadcrumb header. Customize panel with checkboxes and S/M/L size badges. All widgets progressively collapse content based on container height.
Resource usage dashboard widget — CPU and memory usage bars with top-N resource consumers, progressive detail at different widget sizes.
Fleet-aggregate stats subsystem (commits feature/v1.5-rc17). New ContainerStatsAggregator polls watched local and agent-owned containers once per tick (default 10 s) and computes a fleet-wide ContainerStatsSummary (total CPU%, total memory, top-N rows). Two new endpoints — GET /api/v1/stats/summary and GET /api/v1/stats/summary/stream — expose the current snapshot and a live SSE feed; the dashboard Resource Usage widget now consumes the SSE stream directly, fixing the regression (introduced in rc.13 by the ?touch=false workaround) where the widget showed zeros because the per-container cache was never warmed. The legacy GET /api/v1/containers/stats endpoint and the client-side summarizeContainerResourceUsage rollup have been removed.
Per-container update locks (commit 761fb834). New keyed LockManager primitive in app/updates/lock-primitives.ts replaces the module-level pLimit(1) that was serialising every container update across the entire process. Lock keys are derived per container (and per compose project for Dockercompose), so two unrelated containers can now pull and recreate concurrently while two services in the same compose project still serialise correctly.
Restart recovery for queued and pulling updates (commit 00788b13). Startup reconciliation in app/store/update-operation.ts is now selective: status=queued operations stay queued for the recovery dispatcher to pick up, and phase=pulling rows are reset to queued (pull is idempotent). A new app/updates/recovery.ts module runs once after registry.init(), re-resolves trigger and container for each queued operation, and dispatches them through the existing fire-and-forget pipeline.
Notification outbox with retry and dead-letter queue (commits a9561d93, 7d2ef6eb, b215d295, ce26bece). New notificationOutbox LokiJS collection and app/notifications/outbox-worker.ts background worker provide durable retry semantics for notification dispatch. On failure, the delivery intent is persisted to the outbox and the worker retries on a periodic drain with exponential backoff + jitter. After a configurable number of failed attempts (default 5) entries transition to the dead-letter queue; delivered and dead-letter entries are auto-purged past TTL (default 30 days). New /api/notifications/outbox REST surface lets operators list entries, retry from the DLQ, or discard.
Notification outbox UI (commit feature/v1.5-rc17). New Notification outbox page (route /notifications/outbox, nav under Settings) with status tabs (Dead-letter / Pending / Delivered), retry and discard actions.
Cancel queued or in-flight updates (commits 4b79e3ac, 79487115). POST /api/operations/:id/cancel now accepts both queued and in-progress operations. Queued ops are marked failed immediately; in-progress ops are flagged via a cancelRequested field and the lifecycle observes the flag at three safe checkpoints.
Global concurrent-update cap (DD_UPDATE_MAX_CONCURRENT). New counting semaphore provides a configurable global gate on how many update lifecycles run simultaneously. Default 0 = unlimited. Positive integer N means at most N updates run concurrently. Self-update operations bypass the global cap.
Health-gate SSE heartbeat (DD_UPDATE_HEALTH_GATE_HEARTBEAT_MS). While drydock waits for a new container to pass its health gate, a periodic heartbeat re-emits phase: 'health-gate' at a configurable interval (default 10 s). DD_UPDATE_HEALTH_GATE_HEARTBEAT_MS=0 disables heartbeats; values below 1000 ms or non-integers fail fast at startup.
Post-start liveness grace window (DD_UPDATE_POST_START_LIVENESS_GRACE_MS). After Docker start() returns, Drydock waits this many milliseconds and then re-inspects the new container. If the container has already exited, the lifecycle throws and the existing rollback machinery takes over — catching containers that exit immediately after an update (bad command, broken entrypoint, missing dependency) that would otherwise be recorded as a successful update. Default 2000 ms. Set to 0 to disable the check entirely. Values between 1 and 99 ms are rejected at startup; the minimum non-zero value is 100 ms.
Recovery-boot concurrency cap (DD_UPDATE_RECOVERY_BOOT_CONCURRENCY). When Drydock restarts after a crash it finds queued update operations left from the previous run and resumes them. This variable bounds how many are dispatched in parallel during that recovery sweep. Default 4. Values of 0 are rejected at startup (minimum is 1).
Self-update now works when Drydock reaches the Docker daemon over a TCP host, not only through a bind-mounted /var/run/docker.sock (commit fc34ffb9). resolveHelperDockerConnection now inspects the watcher's Dockerode connection: a TCP host produces a TCP helper attached to Drydock's own Docker network. The bind-mounted-socket path is unchanged.
The per-container Update button is locked with a Self-update unavailable indicator when Drydock cannot update itself in the current deployment (commit cf777280). A new hard self-update-unavailable update-eligibility blocker is raised when self-update cannot run over either a bind-mounted socket or a TCP host.
i18n coverage extended to the remaining hardcoded UI strings across 28 components (discussion #329). All 16 non-English locales now have full key parity with the English source. 17 locales ship in the picker: de, es, fr, it, nl, pl, pt-BR, tr, zh-CN, zh-TW, ar, ja, ko, ru, uk, vi, plus English.
DD_AGENT_ALLOW_INSECURE_SECRET escape hatch for closed-LAN deployments. rc.20 tightened the agent-secret-over-HTTP check to a hard error. rc.21 introduces DD_AGENT_ALLOW_INSECURE_SECRET=true as an explicit controller-side opt-in for environments where the operator accepts that the agent secret travels in cleartext. Default behavior is unchanged.
Security scan digest mode. Every scan cycle now carries a stable cycleId (UUID v7) and emits a security-scan-cycle-complete event. Triggers can configure SECURITYMODE=digest (or batch+digest) to receive one summary per cycle. Templates are customizable via SECURITYDIGESTTITLE / SECURITYDIGESTBODY. (#300)
Opt-in scheduled-scan notifications — New DD_SECURITY_SCAN_NOTIFICATIONS=true flag enables security-alert event emission from scheduled scans. Default is false; on-demand scans always emit.
Bulk security scan endpoint — POST /api/v1/containers/scan-all scans all (or a filtered subset of) watched containers server-side, streams per-container progress over the existing scan SSE channel, and honors client-disconnect aborts. Rate-limited to 1 request / 60s per IP (authenticated-admin bypass).
SSE Last-Event-ID replay (#289) — The server stamps every broadcast event with a monotonic <bootId>:<counter> id and retains a 5-minute time-bounded ring buffer. Clients reconnecting with a Last-Event-ID header receive every event they missed; if the buffer has evicted the requested id the client receives a dd:resync-required event.
Update-eligibility blockers on container rows — Backend surfaces 12 structured blocker reasons per container, rendered inline on the Containers list so users see why a row isn't updating without opening the detail drawer.
GET /update-operations/:id endpoint — Returns the current state of a specific update operation for reconciliation when the terminal SSE is missed.
Inline update action in Security view (#299) — Image rows in the Security view now show an "Update" action button directly next to the vulnerability data when a newer image is available.
Watcher next-run metadata (#288) — Watcher API and Agents view now show when each watcher will next poll for updates, with an absolute-timestamp tooltip on hover.
Backend-driven update queue — Container updates are now queued server-side with per-trigger concurrency limits. UI shows Queued → Updating → Updated state progression with sequence labels (e.g. "Updating 1 of 3").
Registry 429 / 503 retry with Retry-After and per-host token bucket (commit ffd1b57b, #342). A new withRetry helper wraps every registry HTTP call: on 429 or 503 it honors the upstream Retry-After header, then falls back to exponential backoff (1 s / 2 s / 4 s, capped at 60 s), up to 3 retries. A new per-host token bucket prevents the watcher from self-inflicting rate limits during a large cron cycle.
Release notes inline popover (commit 09475fa6). The release-notes icon on container rows now opens an inline popover showing both the current and the available-version release notes side by side, with expand/collapse per panel.
Container source project shortcut link (Discussion #295) — Containers now render a clickable "View project" link next to release notes when an org.opencontainers.image.source OCI label, dd.source.repo override label, or GHCR-derived source URL is available.
Actionable deprecation banners (Discussion #214) — The 5 deprecation warning banners now show the concrete migration action inline and include a "View migration guide" link that deep-jumps to the relevant anchored section of the deprecations docs page.
Notification dropdown rework + themeable zebra stripes (Discussion #267) — Header carries the "Notifications" title plus a "Clear" text button, each row shows a per-entry dismiss affordance, and a split footer exposes "Mark all as read" + "Open audit log". Introduces --dd-zebra-stripe, a new theme token.
Notification history store — New LokiJS collection (notifications_history) records a per-(trigger, container, event-kind) result hash so once=true dedup survives process restarts.
Floating tag detection and UI indicator — New tagPrecision classifier (specific | floating) detects mutable version aliases and auto-enables digest watching on non-Docker Hub registries. Container detail views show a caution badge when a floating tag is detected without digest watching enabled. (Discussion #178)
Hide Pinned containers toggle — Checkbox in the container list filter bar hides containers pinned to specific versions. Persisted in user preferences. (Discussion #250)
Combined batch+digest notification mode — Triggers can now use MODE=batch+digest to send both immediate batch emails and scheduled digest summaries. (#254)
Multi-select event-type filter in audit log (commit 5e2d0c70, Discussion #332). The audit log's event-type filter is now a checkbox dropdown supporting any combination of event categories simultaneously.
Bearer token auth for /metrics endpoint — Set DD_SERVER_METRICS_TOKEN to authenticate Prometheus scrapers via Authorization: Bearer <token>.
Disable default local watcher — Set DD_LOCAL_WATCHER=false to prevent the built-in Docker watcher from starting, useful for controller-only nodes that manage remote agents exclusively.
Multi-server notification identification (#283) — Notifications automatically include a [server-name] prefix when agents are registered. Controller name configurable via DD_SERVER_NAME. Custom templates can use container.notificationServerName and container.notificationAgentPrefix.
Infrastructure update mode — dd.update.mode=infrastructure label for socket proxy containers enables helper-swap update path bypassing the socket proxy.
i18n framework migration (refs #329). Bulk vue-i18n migration into per-namespace JSON catalogs under ui/src/locales/en/ (eight namespaces auto-loaded by import.meta.glob in boot/i18n.ts). Foundation for the Crowdin integration. 17 locales ship in the picker.
Design system components — Added shared UI building blocks: AppIconButton, AppBadge, StatusDot, DetailField, and AppTabBar. (Discussion #199)
Podman API version negotiation — Docker watcher probes the daemon's /version endpoint over the Unix socket and pins Dockerode to the reported API version. Prevents EAI_AGAIN crashes caused by docker-modem's redirect-following bug when Podman returns HTTP 301 for unversioned API paths. (#182)
System log live streaming in UI — Added end-to-end WebSocket support for system logs (/api/v1/log/stream) with new UI service/composable and live log view integration.
System log viewer overhaul — Toolbar stays pinned at top, long lines wrap at viewport width, search matches component/level/channel fields, filter toggle shows only matching entries, sort toggle switches between oldest-first and newest-first. (#259, #260)
Rollback shortcut in container actions menu — Quick rollback option directly from the container row actions dropdown.
SPA + hashed-asset cache-control — Static UI assets with hashed filenames are served with immutable long-lived cache headers; the SPA index.html carries a short revalidation header.

Changed

Legacy aggregate container stats route now fails explicitly. GET /api/v1/containers/stats is still removed from the public stats API, but the router now catches the legacy path before /:id and returns 410 Gone with migration targets (GET /api/v1/stats/summary for aggregate stats, GET /api/v1/containers/:id/stats for per-container stats) instead of treating stats as a container id.
Default watcher cron relaxed from hourly to every 6 hours (#342 follow-up). app/watchers/providers/docker/Docker.ts now defaults cron to 0 */6 * * * (every 6 hours) instead of 0 * * * * (hourly). Users who set DD_WATCHER_{name}_CRON explicitly are unaffected. Users who want near-real-time detection can still set DD_WATCHER_{name}_CRON=0 * * * *.
Action trigger default mode — Action triggers (docker, dockercompose, command) now default to AUTO=oninclude instead of AUTO=all, requiring an explicit dd.action.include label before auto-updating containers. (#213)
Self-update helper now prefers the bind-mounted Docker socket over a TCP watcher connection (commit aa828d88). The resolution order is now inverted: findDockerSocketBind runs first, and if the target container carries a socket bind the helper uses that direct socket path regardless of the watcher's TCP configuration. TCP is the fallback for pure socket-less deployments.
DD_SESSION_SECRET auto-generated and persisted when unset. On first boot without DD_SESSION_SECRET set, drydock generates 64 random bytes and writes them to a secrets collection inside /store/dd.json. Subsequent boots read the persisted value so sessions survive restarts. The env var still takes precedence when set. (rc.21 restored this after rc.20 made it a hard requirement without a migration path; existing deployments that set the variable see no change.)
Watcher dispatch is fully fire-and-forget (commit 5cfa2286). Trigger.runUpdateAvailableSimpleTrigger and runAcceptedUpdateBatch no longer await runAcceptedContainerUpdates, so a slow update lifecycle no longer stalls the next watcher tick.
"Update started" toasts renamed to "Update queued" (commit 79487115). Dispatch is fire-and-forget — the text now matches what actually happened.
Shared DataTable column sizing overhaul (commit 596adcd2). All first-party table surfaces now route through the shared DataTable component with numeric sizing metadata, supporting pointer and keyboard column resizing, double-click autosize, and persistent manual/autosized widths per table.
Crowdin export configuration aligned with app locale folders.
Tag-last release pipeline (fixes #306). Collapsed release-cut.yml and release-from-tag.yml into a single workflow where the git tag push is the last step, performed only after the Docker image has been built, pushed, signed with cosign, attested (SLSA), and the signed release tarball has been verified. Enforces the invariant: if the git tag exists, the image exists.
Playwright E2E tests moved to a dedicated workflow file (e2e-playwright.yml) (commit f0989301). OSSF Scorecard's CI-Tests check now scores independently from the main ci-verify suite.
Translations refreshed from Crowdin (commit 202f3d83). Human translations synced from Crowdin for the rc.23 i18n extraction sweep, updating the 16 non-English locales across all UI namespaces.
Security alert emit is non-blocking inside the update lifecycle (commit 6c5198dd). SecurityGate.maybeEmitHighSeverityAlert now returns synchronously after firing the emit; the lifecycle no longer waits for sequential provider notifications.
Expand all / Collapse all bulk toggle — Replaced the single chevron toggle in the Containers toolbar with an explicit "Expand all" / "Collapse all" button.
Soft eligibility blockers de-emphasized by default (Discussion #325). Soft blockers now render with neutral muted styling so hard blockers visually dominate the row, and active blockers sort hard-first.
Responsive dashboard layout persistence — Dashboard widget bounds and layout are now breakpoint-aware, persisting separate layouts per viewport tier.
Trigger digest flush DRY refactor — flushDigestBuffer / shouldHandleDigestContainerReport are now parameterized on the event kind so update-digest and security-digest share a single implementation.
Healthcheck execution path optimized — Default HEALTHCHECK probe replaced with a 65KB static C binary (/bin/healthcheck). curl is retained for backward compatibility with user-defined HEALTHCHECK overrides during the deprecation window (scheduled for removal in v1.7.0, final warning release v1.6.0).
Agent reconnect notification — New opt-in agent-reconnect notification rule that fires when a remote agent reconnects after losing connection. Disabled by default.
app/updates/locks.ts renamed to app/updates/lock-primitives.ts (commit 4c506d21). The module now contains general-purpose synchronisation primitives (Semaphore, LockManager) not tied to the updates subsystem.
Exact-version package.json pinning — Flipped the four remaining caret specifiers to exact versions so every package.json matches the already-exact lockfile resolutions. Every dependency layer is now SHA-immutable.
Dashboard reconnect refresh is live-only. On dd:sse-connected the dashboard now refetches only endpoints that can go stale between frames; dd:sse-resync-required still forces a full fan-out.

Deprecated

DD_TRIGGER_* environment variable prefix and dd.trigger.* container labels (deprecated v1.5.0, removal targeted v1.7.0). Use DD_ACTION_* / dd.action.* for update-action triggers (Docker, Docker Compose, Command) and DD_NOTIFICATION_* / dd.notification.* for messaging/notification triggers (Slack, SMTP, Discord, Telegram, ntfy, Pushover, and all others). The legacy prefixes continue to work as aliases through v1.7.0. A migration CLI (drydock config migrate --source trigger) rewrites existing configs automatically. See deprecations for the full schedule.
dd.action.include / dd.action.exclude (and legacy dd.trigger.include / dd.trigger.exclude) become hard manual-update blockers in v1.7.0. v1.5.x keeps them as soft blockers — the pill reads Trigger filtered / Trigger excluded but the manual Update button stays clickable (with a warn-and-confirm). v1.7.0 will lock the button and reject the API call when the labels filter out the matching trigger. See DEPRECATIONS.md for migration guidance.
curl healthcheck override — curl is retained in the image for user-defined HEALTHCHECK overrides during the v1.5.x deprecation window; removal is scheduled for v1.7.0 (v1.6.0 is the final warning release).
Agent-less Home Assistant MQTT topic layout for multi-agent deployments (deprecated v1.5.0, default flips v1.7.0). When more than one node uses the default watcher name local, the current agent-less topic layout makes same-named containers on different agents publish to and overwrite the same MQTT topics, including watcher running-status topics. Set DD_NOTIFICATION_MQTT_<name>_HASS_AGENTTOPICSEGMENT=true to opt into the corrected per-agent layout now; it becomes the default in v1.7.0. Single-node deployments are unaffected. Enabling it changes the Home Assistant entity IDs for agent-owned containers — see DEPRECATIONS.md for migration guidance.

Removed

Experimental eligibilityPills.{showSoft,deemphasizeSoft} preferences dropped before release. The de-emphasis is now baseline behavior with no toggle.
Experimental containers.showAutoUpdateDiagnostic preference + compact variant of UpdateEligibilityBadges dropped before release.
Two pre-existing unused imports flagged by biome's noUnusedImports (updates/request-update.ts default Trigger import; Docker.containers.processing-retrieval.test.ts mockGetFullReleaseNotesForContainer import).
Legacy GET /api/v1/containers/stats endpoint and summarizeContainerResourceUsage client-side rollup removed; superseded by the new fleet-aggregate stats subsystem (GET /api/v1/stats/summary).

Fixed

#418 — Operators behind a TLS-terminating reverse proxy now get an actionable startup log when manual updates fail with 403 CSRF validation failed. The rc.30 same-origin hardening made DD_SERVER_TRUSTPROXY mandatory behind a proxy that terminates TLS (without it, drydock sees req.protocol as http while the browser sends an https Origin, so every state-changing request is rejected) — but the only symptom was an opaque 403 with no pointer to the fix. When trust proxy is disabled and a request arrives carrying X-Forwarded-Proto: https, drydock now logs a one-time warning naming the exact cause and the DD_SERVER_TRUSTPROXY=1 fix (linking the reverse-proxy FAQ). The CSRF guard itself is unchanged — genuine cross-site forgery is still rejected.
#415 — Registries that require a Bearer token exchange for tag lookups (Chainguard cgr.dev, Codeberg/Forgejo codeberg.org, and other standard Docker v2 registries) no longer fail with 401 Unauthorized. callRegistry now implements the spec-compliant flow: on a 401 carrying a WWW-Authenticate: Bearer realm=…,service=…,scope=… challenge it fetches a token from the realm (anonymously for public images, or with the configured credentials) and retries the request once. The realm host is validated against the registry's own host via the existing validateAuthUrlHost guard, and the handler falls back to the original error if the exchange can't be completed, so it is purely additive for providers that already authenticate. Fixes tag enumeration for public Codeberg images and credentialed Chainguard images.
#413 — Containers deployed as a Docker Swarm stack (e.g. Portainer agents via docker stack deploy) now group in the container stack view. The grouping key derivation in app/api/group.ts recognized com.docker.compose.project but not com.docker.stack.namespace, the label Swarm stacks carry instead of the Compose project label — so Swarm-deployed services fell through to the ungrouped bucket while Compose services grouped correctly. com.docker.stack.namespace is now the lowest-priority auto-detected group key, after dd.group > wud.group > com.docker.compose.project.
#411 — Multi-agent deployments no longer produce a false 409 Conflict when identically-named containers are being updated on different agents. The duplicate-update check revived in rc.30 fell back to a global name-only lookup (getActiveOperationByContainerName) with no agent+watcher scope. Containers on different hosts are physically separate — the lookup is now scoped by { agent, watcher } via an optional ContainerIdentityFilter parameter added to both getActiveOperationByContainerName and getInProgressOperationByContainerName in app/store/update-operation.ts. Legacy operation rows that predate the container-snapshot field (#385) are accepted unconditionally as a backward-compatible fallback. The same scoping is applied to the ContainerUpdateExecutor reconciliation path, the list-handler eligibility context, and the SSE container-enrichment path so cross-agent contamination is closed at all five call sites.
Latent-bug audit of the 1.5 RC bug classes — ten same-class defects fixed before 1.5.0 final. A class-wide sweep of the recurring root-cause classes behind the 1.5 RC churn (container-identity-by-name, missing agent scope, terminal-event snapshots, operationId threading, notification dedup, stale-scan epochs, SSE reconnect, image-reference normalization, health readiness) surfaced the following still-unfixed instances. Each is active by default and ships with a regression test:
- Agent-hosted update operations now carry a container snapshot and can no longer be orphaned mid-prepare. ContainerUpdateExecutor created fresh operation rows without a container, so terminal update-failed/rollback events emitted with no snapshot and the notification fell back to a store lookup that misses after a recreate — silently dropping the agent-path failure notification. The post-pull tail (clone-runtime-config + rename) was also unguarded, leaving the row stuck in-progress for the full TTL on error. The snapshot is now persisted at enqueue and the tail marks the operation terminal on failure. Latent remnant of the #385/#386 terminal-lifecycle class.
- The direct remote-trigger route (POST /:type/:name/:agent) now threads the controller-minted operation id (#289 remnant). For update triggers it creates the controller-side queued row, honors a caller-supplied operationId (previously validated then silently discarded), forwards it to the agent as runtime context, and returns 202 { operationId } instead of 200 {}; UpdateRequestError is surfaced with its status. The agent no longer mints a divergent id the controller never tracks.
- Digest- and batch-mode notifications no longer re-fire a spurious "update available" after a successful update (#408 digest/batch remnant). The original #408 guard (recentlyAppliedContainerKeys) was applied only to the simple report path, so a stale updateAvailable=true report arriving before the watcher caught up could still re-buffer a just-updated container on the digest and batch paths. The digest path (shouldHandleDigestContainerReport plus the handleContainerReportDigest lift) and the batch path (shouldHandleBatchContainerReport plus a new handleContainerReports lift) now share the same suppress-then-lift lifecycle. The batch-path lift is mandatory rather than belt-and-braces: a pure mode=batch trigger registers neither the simple nor the digest handler, so without it the suppression key would never clear and that container's update-available notifications would be muted permanently after the first successful update.
- Re-armed the post-update stale-scan guard (#265 regression). The rc.17 operation-store refactor removed the only call that stamped the manual-update epoch, leaving preserveClearedUpdateState as dead code; an in-flight cron scan could re-raise the cleared update badge. maybeFastResyncAfterUpdate now stamps the epoch before the resync scan, suppressing earlier in-flight scans.
- The Security view now refreshes its container list on SSE reconnect/resync. It previously refetched only vulnerabilities, leaving update-eligibility, blocker tooltips, and the inline Update action stale after a dropped connection (it could offer Update on a container that no longer exists).
- The self-update helper image reference is now built with buildImageReference. resolveHelperImage hand-rolled the registry-URL normalization; when registry.url ended with a trailing slash (…/v2/) the concatenation produced a double-slash reference (ghcr.io//org/image:tag) that Docker rejects with HTTP 400. It now delegates to the canonical helper that strips the scheme and trailing /v2[/] before concatenation.
- Scheduled security scans write their result onto the current container record, not a stale snapshot. The async Trivy scan back-wrote by spreading the container snapshot captured at batch-prep time, carrying stale update-state and — on the recreate path, where the container has a new id — creating a zombie store record for the old id. The write-back now re-reads the live record by id, skips if the container is gone, and merges only the security field.
- The agent /health endpoint now reflects watcher-registration failure. It was a hardcoded 200, so an agent whose watchers all failed to register (unreachable Docker socket, bad TLS — swallowed with log.warn) reported healthy forever. It now returns 503 when zero watchers are registered, mirroring the main API's readiness gate; the normal (≥1 watcher) path is unchanged.
- The agent batch-completed event parser resolves controller-issued operation ids. parseBatchUpdateCompletedPayload was the only event parser scoping its operationId with toAgentScopedId instead of resolveAgentOperationId; a controller-owned id reaching it would be double-scoped and stop matching the controller's row. It now resolves consistently with the other parsers (#289 class).
- Container-recreate no longer fails with HTTP 403 on deployments using a hardened socket proxy that enforces a runtime allowlist (commit af79e612). Drydock copied the daemon-reported HostConfig.Runtime ("runc") verbatim into the POST /containers/create body, and a socket proxy with an explicit runtime allowlist rejected it as not-allowlisted — even though the container was only ever using the daemon default. The fix fetches the daemon's DefaultRuntime via GET /info during the clone-prepare step and omits HostConfig.Runtime from the create body when it merely restates that default. Explicitly-selected non-default runtimes (nvidia, kata, sysbox-runc) are preserved, and when /info is unavailable the runtime is left untouched (prior behaviour). The HostConfig is shallow-cloned before editing so the inspect spec kept for rollback is not mutated.
- Stuck update operations now terminalise to a non-notifying expired status instead of failed (#410). The active-TTL sweep and the startup-orphan reconciliation marked timed-out queued/in-progress rows as failed, which fired markOperationTerminal's update-failed lifecycle event — a false "update failed" notification that hit two distinct cases: a genuinely orphaned operation (an agent-scoped controller trigger that errored after the queued row was created but before the agent ran it) and a slow-but-successful update whose agent confirmation simply outran the 30-minute TTL. Both now resolve to a new terminal status, expired, for which emitTerminalLifecycleEvent emits nothing at all, so neither case can cry wolf. The UI treats expired as terminal — the "Updating" badge clears with no failure styling and no toast. A late agent report arriving after expiry is ignored (the row is already terminal), so such an operation reads as expired rather than flipping to a delayed success/failure. A second, deeper instance of the same false alarm is now closed: when a duplicate update operation executes against a container that an earlier pass — or Docker Compose, or an agent — already recreated, the rename/refresh hits a genuine Docker 404 (no such container / no longer exists) or a 409 conflict (a real error, not a TTL timeout), which previously terminalised as failed and fired the same ghost "update failed" notification. The duplicate-op path now defuses these benign post-success errors at three choke points: the dead name-based dedup fallback in getActiveUpdateOperationForContainer is repaired so a stale-container-id duplicate is blocked up front (409 instead of silently slipping through); and both ContainerUpdateExecutor's rename() catch and the Docker lifecycle outer catch consult a shared duplicate-op-classification helper that downgrades failed→expired only when a succeeded terminal row already exists for the same container name inside a 15-minute window. A real failure with no recent success still terminalises as failed and notifies. Terminal-lifecycle/orphan class, sibling to #385/#386.
- The fleet-stats collector now backs off on stream reconnect instead of hot-looping. When a container's Docker stats stream closed immediately on open (an exited container, a daemon error, or a malformed stream), restartCollection re-opened it synchronously — a tight reconnect loop bounded only by a single microtask tick that spins CPU and floods logs, once per such container. Reconnects now use capped exponential backoff (1 s doubling to a 60 s ceiling), reset to the base delay once the stream delivers data; the scheduler refuses to stack timers, skips reconnecting once a container is unwatched, and clears the pending timer on the last release. Latent defect in the new-in-1.5 fleet-stats subsystem.
#386 — Agents permanently showing 0 running containers in the controller UI (the recurrence that survived the rc.20/25/26/28 fixes). Root cause was two compounding bugs outside the handshake/snapshot machinery the earlier fixes targeted. (1) The controller's own local Docker watcher was pruning every agent's containers. Its store query getContainers({ watcher: this.name }) was scoped by watcher name only, and remote agent containers are stored under the same default watcher name (local), so each controller watch cycle treated the agents' rows as stale (their IDs are not present on the local Docker), failed to inspect() them, and deleted them — roughly every 6h until the agent was restarted. This is also why the rc.28 handshake-0 guard appeared not to work: the rows were wiped seconds before the handshake ran, so there was no "last-known state" left to preserve. Docker.getContainers now scopes containersFromTheStore by normalizeAgentValue(this.agent), so a watcher only ever prunes its own agent's containers. (2) A lost dd:watcher-snapshot was never recovered until the next 6h cron. The agent emits its authoritative snapshot only at the end of each cron; if the controller's SSE stream was silently half-open or mid-reconnect at that instant the snapshot was written to a dead socket and lost, leaving the controller empty. The agent now caches the latest snapshot per watcher and replays it to each new SSE client immediately after dd:ack, so the controller converges on the agent's true container set on any (re)connect. This also explains why agents with many containers were immune (constant per-container SSE traffic kept the connection warm) while low-traffic agents went idle and dropped.
#386 follow-up — six more cross-agent contamination bugs of the same class, found by a class-wide audit. The prune/snapshot fix above closed the headline symptom; an audit of every store read/key/match that scoped by watcher name or container name without also scoping by agent (the controller's own local containers and a remote agent's local containers share the default watcher name) found and fixed the following further sites. These are active by default:
- Post-update fast-resync (Docker.maybeFastResyncAfterUpdate) matched the resync target by watcher_name and could pick a different agent's same-named container, leaving a stale "update available" badge until the next cron — the candidate set is now agent-scoped.
- Deferred update reconciliation (scheduleDeferredReconciliation) resolved the container by name only and could no-op against another agent's same-named row, leaving the local operation stuck in rollback-deferred until restart — it now resolves by operation/container id with an agent-scoped name fallback.
- Watcher-list stats (GET /api/watchers, GET /api/watchers/:id) keyed the per-watcher stat buckets by bare watcher name, collapsing every local watcher into a single inflated bucket — they are now keyed by the unique registry id (docker.local vs ml.docker.local), so each watcher row reports its own counts.
- Dashboard stats aggregator (stats/aggregator.ts) mapped agent containers to the controller's local watcher, silently dropping them from the fleet summary (or misattributing CPU/memory on a name collision) — it now resolves the agent-aware watcher id, so agent containers are no longer queried against the wrong Docker socket.
- Security-scan result cache (store/container.insertContainer) keyed cached Trivy results by watcher_name only, so a remote agent's same-named container could be stamped with the controller's scan verdict (false-clean / false-alarm) — the cache (only ever written for controller-local containers) is now skipped for agent containers.
- Name-based webhook routing (POST /watch/:name, POST /update/:name) resolved the container by scanning the entire store by name and fanned the action out to all watchers; it now dispatches only to the resolved owning watcher and returns 409 Conflict when a name is ambiguous across agents/watchers (disambiguate with ?agent= / ?watcher=). Single-match (single-host) behavior is unchanged.
The Home Assistant facet of this class (watcher-level sensor counts summing across agents, and discovery-entity cleanup leaking ghost entities) is corrected by the new opt-in DD_NOTIFICATION_MQTT_<name>_HASS_AGENTTOPICSEGMENT flag, because closing it requires a change to the MQTT topic structure — see Added and Deprecated.
The Docker image builds again after Alpine edge/testing dropped the pinned trivy=0.70.0-r1 package. apk add failed with breaks: world[trivy=0.70.0-r1], which broke the multi-arch image build (and the Cucumber E2E job that builds it). trivy is now pinned to the package name only, matching the existing curl exception in the same layer: edge/testing rotates and drops -rN builds quickly and per-arch mirrors desync during the window, so an exact -rN pin is inherently fragile for multi-arch builds.
#408 — A successful update no longer fires a spurious "new version available" notification. handleContainerUpdateAppliedEvent clears the once notification history after an update is applied, which re-opens the once gate before the watcher's next scan sets updateAvailable=false. A lingering container report still carrying updateAvailable=true (common during "Update All") then fired a duplicate "update available" message right after "updated successfully". A per-container suppression set now withholds update-available notifications for a just-applied container until a watcher report confirms updateAvailable=false, after which the key is lifted so future real updates still notify. Additionally, update-applied and update-failed lifecycle notifications now bypass the semver threshold filter, so a post-update updateKind.kind of unknown (e.g. digest-only updates) under a non-all threshold no longer silently drops the success/failure notice.
#391 — A failed Docker Compose update no longer destroys the running container. refreshComposeServiceWithDockerApi previously removed the old container before recreating it, so any failure of the create step (e.g. the newly-pulled image is not available for the host platform) left the service with zero containers and no recovery. The update path now (1) performs a pre-flight check that the pulled image is architecture-compatible with the host and aborts before removing the old container if it is not, and (2) wraps the stop/remove → recreate sequence so that, if recreate still fails, the original container is restored from its captured spec and the original error is re-thrown. The running container is never lost on a failed update.
#402 — /health now returns 503 until passport authentication strategies are fully registered, closing a startup race where the port accepted connections before auth was ready (intermittent 401s on startup). registerRoutes() in the auth subsystem now calls setAuthReadyFn before auth.init(), wiring the readiness gate before strategy registration begins. Deployments that hit intermittent 401 errors during the first seconds after container start should see the issue resolved.
#386 — A fresh-restart agent whose in-memory store has not yet been re-populated no longer wipes the controller's last-known container state on handshake. AgentClient._doHandshake now skips pruneOldContainers whenever containers.length === 0 (and hasConnectedOnce is true) and emits a warning. Pruning is deferred to the next authoritative dd:watcher-snapshot, which is already gated on !containerEnumerationFailed && enrichmentErrors === 0 and is therefore unambiguous.
#386 — Agents intermittently showing 0 running containers in the controller UI (multiple recurrences across rc.20–rc.26). The complete fix spans three layers: (1) rc.20 introduced containerEnumerationFailed in Docker.watch() to suppress snapshots when getContainers() throws; (2) rc.25 extended suppression to per-container enrichment errors via a diagnostics.enrichmentErrors out-parameter; (3) rc.26 added a dedicated AgentStatsChanged event so the controller UI also refreshes on Docker-event-driven container changes (starts/stops between cron cycles), not only completed cron cycles.
#289 — Agent-hosted container updates no longer leave an orphaned queued operation row on the controller that the 30-minute TTL sweep force-fails into a misleading "update failed" notification. The fix threads the controller's operationId end-to-end: AgentTrigger.trigger / triggerBatch now accept and forward runtimeContext; AgentClient.runRemoteTrigger extracts per-container operationIds and includes them in the agent payload; the agent-side controller runTrigger accepts and threads the operationId into requestContainerUpdate; a new AgentClient.resolveAgentOperationId helper reuses the controller-side row when found. The controller-side queued row therefore transitions directly to in-progress and succeeded/failed from the agent's lifecycle events, eliminating the spurious "update failed" notification.
#289 — Update-applied and update-failed notification triggers and UI success toasts no longer silently drop for containers running on a connected agent. The agent's container snapshot is now threaded through buildAgentOperationBase, ensureAgentOperationForTerminal, markAgentOperationTerminal, and related helpers, stamping agent: this.name. The store's terminal-lifecycle emit therefore naturally carries the agent's container into emitContainerUpdateApplied / emitContainerUpdateFailed, and both the notification trigger and the SSE toast fire end-to-end on the controller for agent-originated updates.
#289 — Container rows no longer drop sort position during an in-flight update, and every terminal outcome (succeeded / failed / rolled-back) now fires a toast. The operation display hold captures a sort-field snapshot at hold start so the row stays pinned through the docker recreate window.
#290 — "Updated Successfully" toast no longer drops intermittently after a container update. A new useGlobalUpdateToast composable mounted once at App.vue is the single source of truth: listens for dd:sse-update-applied / dd:sse-update-failed / dd:sse-batch-update-completed, survives route navigation, dedupes by operationId over a 5-minute window, and waits for the matching dd:sse-container-added/updated/removed event before firing.
#291 — Dashboard update flow now fires the same toast sequence as the Containers view and shares the same useOperationDisplayHold composable, fixing the last reporter symptoms (updating row no longer drops to the bottom mid-update; dropped terminal SSE no longer leaves the dashboard silent).
#342 — A container is no longer shown as "update available" with a blank target version after a transient registry error. hasRawUpdate in app/model/container.ts now performs the tag comparison only when both image.tag.value and result.tag are defined, matching the existing guard in getRawTagUpdate.
#342 — GitHub release-notes lookups now survive GitHub's secondary rate limit instead of giving up on the first burst. GithubProvider classifies a 403 as a secondary rate limit only when it carries a retry-after header or x-ratelimit-remaining: 0, retries those, and arms a short module-level cooldown driven by GitHub's own retry hint. The withRetry helper gains optional retryPredicate and retryDelayMs hooks.
#342 — Registry routing always uses the credentialed instance when one is configured (commit 069274fe). The router now gives explicit priority to credentialed instances. The silent anonymous fallback on 401/403 is also removed; auth rejection now throws an actionable error that surfaces as the "Check failed" badge in the UI.
#342 — The registry-error tooltip on the Containers view now names the registry that failed. registryErrorTooltip now derives the registry hostname from the container's registryUrl and renders it through a new i18n string, e.g. ghcr.io — Request failed with status code 429.
#342 — Hybrid image:tag@sha256:digest refs no longer trigger a spurious "Cannot get a reliable tag" warning when Docker's RepoTags is empty. Docker.resolveImageName now detects hybrid refs and parses them directly via parse-docker-image-name; only true digest-only refs fall through to the existing resolveDigestOnlyImage path.
#342 — Containers list "Version" column shows the correct data for all container types. Digest-pinned containers continue to show the sha256:abc… → sha256:def… pair; floating-tag + digest-watch containers render the human-readable tag with the digest delta in the tooltip; hybrid digest containers show the tag with the digest pair inline.
#355 — update-failed notifications no longer drop silently when the controller's container store races against post-failure prune. UpdateLifecycleExecutor now carries the failing container on the update-failed payload, and Trigger.handleContainerUpdateFailedEvent accepts payload.container as the primary source with the store lookup as fallback.
#357 — Transient Trivy failures no longer wipe previously-stored scan history. The scheduler now keeps the existing record when the new result is an error and the existing record is less than 7 days old. Error results are also no longer indefinitely re-spawning fresh Trivy invocations — scanImageWithDedup uses a 15-minute error retry floor.
#357 / #355 — Trivy scan and SBOM no longer require /var/run/docker.sock inside the drydock container. The forced --image-src docker flag is removed; Trivy now uses its default source order and falls back to a registry pull when the local daemon isn't reachable. Set DD_SECURITY_TRIVY_IMAGE_SRC=remote to skip the local-daemon probe.
#370 — Containers list "Version" column again shows the human-readable image tag for floating-tag + digest-watch containers (rc.20 inadvertently reverted the #356 fix). The updateKind === 'digest' && !isDigestPinned branch has been restored to the correct behavior: CopyableTag with the full digest delta in the cell tooltip.
#371 — Containers "Group By Stack" view no longer dissolves a multi-container stack into "Ungrouped" while its last container is mid-update. A new groupAssignedSizeMap ref records each group's API-assigned member count; the flatten condition now requires both buckets[key].length === 1 and groupAssignedSizeMap.value[key] === 1.
#374 — Security scans no longer hand Trivy a raw registry v2 API URL. The resolveContainerImageFullName fallback now strips the URL scheme and the /v2 path segment, matching Registry.getImageFullName, yielding a plain registry-1.docker.io/image:tag reference.
#385 — Telegram, Pushover, and other notification triggers no longer silently swallow update-applied and update-failed events after a compose recreate or on multi-agent deployments. The fix persists a snapshot of the Container on the operation entry at enqueue time (createAcceptedContainerUpdateRequest) and buildTerminalLifecycleEventBase now forwards that snapshot on the terminal-lifecycle payload. The agent SSE wire was also extended to forward the container snapshot end-to-end.
#328 — Triggering a security scan emitted a "container update" notification instead of the security digest. renderBatchTitle / renderBatchBody / composeBatchMessage now honour runtimeContext.title / .body verbatim when set, so the security-digest path produces the configured securitydigesttitle / securitydigestbody output instead of update-available output.
#317 — A notification trigger configured with auto: false no longer also silently loses lifecycle notifications (update-applied, update-failed, security-alert, agent-connected, agent-disconnected). Auto-fire-on-detection handlers stay gated by auto; lifecycle handlers register unconditionally.
#317 — Update button bypassed eligibility blockers, queuing requests the API would only reject one-by-one. The API now rejects manual updates on any hard blocker with the blocker's user-friendly message; the UI locks the Update button on hard blockers and shows a confirm-modal warning on soft blockers.
#315 — Self-update now works against private registries whose registry.url is stored as the v2 API base (e.g. https://ghcr.io/v2). resolveHelperImage now normalizes the reference to match Registry.getImageFullName.
#308 — Per-row scanning chip is now correctly anchored to the container being scanned and no longer floats in viewport-fixed gutter space. The scan lifecycle uses a useScanLifecycle composable maintaining a scansInFlight set keyed by container id (with a 120s safety timeout). A sibling dd-row-scanning class provides the containing block without dimming the row.
#305 — Hide Pinned now hides every pinned container again, matching rc.8 behavior. (#293 had carved out an exception for pinned rows with a pending update.)
#301 — GET /api/containers preloads all active update operations in a single indexed scan, replacing the rc.8 per-container 3-scan fan-out. Synthetic per-attach time drops from ~5–25 ms to ~0.02–0.04 ms on large inventories.
#296 — Controller identity detection now runs for host-based watchers (TCP to a local socket-proxy, the common Synology / Docker Compose pattern). Set DD_SERVER_NAME to override.
#293 — Hide Pinned filter no longer hides pinned containers that have a pending update. Filter decluttering is preserved for static pinned containers; rows with a pending update (newTag truthy) surface through the filter.
#282 — batch+digest mode now sends both the immediate batch email and the scheduled morning digest for each detected update, matching the documented semantics. The fix splits the digest channel off as its own NotificationEventKind ('update-available-digest') so batch-channel and digest-channel dedup are independent.
#270 — Hide-pinned filter now uses computed tagPinned property instead of stale stored field. Unconditional startup repair ensures tagPrecision data is always correct.
#256 — Update operation state disambiguated by container ID instead of name, preventing cross-host bleed between same-name containers on different hosts.
#253 — Shorthand trigger references resolved in notification rule matching; notification buffering keys stabilized; debug logging added to every silent filter path.
#248 — API guard against duplicate container updates (409 conflict).
#245 — Container update fails with 500 error when no healthcheck — Health gate now skipped when the container has no healthcheck. (This entry originally said the gate was skipped when dd.rollback.auto is not set. That condition was removed before 1.5.0 shipped: the gate applies to any running container with a healthcheck, and dd.rollback.auto only controls monitoring after the update completes.) Pre-healthy timeout uses max(120s, dd.rollback.window) instead of a fixed value.
#238 — Container inspect Config.Image fallback — When Docker summary only exposes a sha256:… image ID (no RepoTags), container discovery falls back to container inspect Config.Image to recover the original tagged reference.
#229 / #228 — Spurious SMTP emails after update — clearDetectedUpdateState() now clears raw result/updateKind data instead of the derived updateAvailable boolean.
#223 — Dashboard layout customizations lost on page reload — Added gridLayout to PreferencesSchema; reorder now uses loadPersistedLayout.
#222 — Dashboard customize panel responsive on mobile — Panel is opt-in on mobile (sliders icon to open), full-screen overlay with backdrop dismiss.
#217 — Dashboard Resource Usage widget minimum height raised from 3 grid units to 7 grid units so per-container CPU and Memory lists stay visible.
#213 — Dashboard Host Status widget no longer auto-scrolls to the last host when the host list changes. The scroll-snap classes, dynamic tail-spacer element, and measurement machinery have all been removed.
#208 — Dashboard updates widget 6-item cap removed — Removed hard-coded RECENT_UPDATES_LIMIT that silently dropped entries beyond 6 in the Updates Available widget.
#202 — CalVer zero-padded month in strict family filter — Tags like 2026.02.0 were rejected when the current tag was 2025.11.1 because zero-padded single digits were treated as a family mismatch.
#200 — Dashboard widget mobile scroll — Added overscroll-contain to all scrollable dashboard widgets.
#192 — Digest-only image visibility — Watchers no longer silently drop containers with digest-only image references.
#186 — Registry failures in Updates Available widget — Containers with "check failed" status no longer appear in the dashboard "Updates Available" section.
#183 — Cascading -old container updates — "Update All" batch no longer triggers updates on containers renamed with -old-{timestamp} suffix during a prior update.
#182 — Podman pod infra containers skipped — Watchers now skip Podman pod infrastructure containers that have an empty Image field.
#180 — Duplicate containers after recreate — Three-layer deduplication filtering prevents alias containers from entering the store during Docker recreate cycles.
#156 — Container alias name canonicalization — getContainerName() now strips Docker recreate alias prefixes (e.g. 8bf70beac570_termix → termix) before the name enters the store, so all triggers receive canonical names. MQTT Home Assistant sensor preserved during recreate (replacementExpected flag prevents premature empty retained discovery payload).
Discussion #295 — Release-notes icon in the container table now always opens the same popover, even when only an external release URL is available. The popover now renders uniformly for both cases.
discussion #295 — DD_SESSION_SECRET no longer crashes startup when unset — The fallback is now a persisted secret: on first boot without DD_SESSION_SECRET set, drydock generates 64 random bytes and writes them to the store. Subsequent boots read the persisted value. (See also the Changed entry above.)
#362 — SSE reconnect exponential backoff no longer collapses to a flat 1 s loop when the agent is struggling. The backoff now only resets after the stream has stayed open for SSE_STABLE_CONNECTION_MS (30 s).
AgentClient timers are now cleared when an agent is removed, preventing orphaned timeouts (commit 03bf7211). A new idempotent stop() method cancels both stableConnectionTimer and reconnectTimer.
#368 — OIDC custom-dispatcher paths (cafile / DD_AUTH_OIDC_*_INSECURE=true) no longer fail with an opaque TypeError: fetch failed on Node 24. The fix imports fetch from undici and uses it whenever a custom dispatcher is required so both halves share the same dispatcher version.
OIDC warn logs now surface the full error.cause chain, making TLS and DNS failures actionable (commit 720d99a3).
OIDC error logs now redact RFC-1918 IP addresses and absolute filesystem paths (commit 9b79de77).
OIDC SSO broken after upgrade/restart — OIDC discovery made lazy so startup failure no longer drops the provider. Redirect, callback, and token paths retry on first use. (#246)
Image reference construction — unanchored /v2 strip could silently corrupt references when the image name contained a /v2 path segment. The fix extracts a shared pure helper buildImageReference (app/registries/image-reference.ts) that cleans the registry URL before concatenation using anchored regexes.
ECR stale auth token cache write avoided on concurrent key change. The cache write is now keyed on the credentials snapshot captured at request start.
Icon proxy serves fallback image on upstream CDN timeout or 5xx. Non-existence failures now route through the existing fallback path and serve the placeholder image.
SBOM endpoint returns 503 instead of 500 when the security scanner is disabled.
Malformed dd.tag.transform regex patterns — The regex-transform label validator now throws at config time for malformed or oversized patterns.
dd.registry.lookup.image label no longer corrupts deploy identity (commit 594a07e8, fixes #336). normalizeContainer no longer overwrites image.name / image.registry.url; a new getImageForRegistryQuery helper applies the substitution only at each query boundary.
Password-manager autofill restored on login form (commit 3abe2fa6, fixes #335). Username and password inputs now carry name and id attributes.
security-scan-skipped audit row now fires when the gate is disabled globally (commit ae24e0a9).
Command action security warning updated to canonical DD_ACTION_COMMAND_* prefix (commit aa5fc98d).
Docker event history pruning amortized to reduce per-event splice cost (commit d6690cc8). The threshold is now 2×maxEntries, so splices are amortized across many appends.
Agent container list no longer shares mutable LokiJS references (commit 1f7d8034). The handler now clones each container via cloneContainer before stripping metadata.
Docker multi-arch build no longer fails when Alpine repos drift between archs. The curl entry is now unpinned so apk installs whatever's current per-arch.
Telegram MarkdownV2 escaping in all trigger paths — Body text in trigger() and triggerBatch() was not escaped for MarkdownV2 reserved characters, causing Telegram API 400 errors and silent notification failures. (Discussion #211)
#310 — Restored the [server] / [agent] prefix on default notification body templates that rc.10 had stripped.
#309 — Status column in the Containers list now shows its label alongside the icon at typical widths.
#296 — Notification server-name prefix no longer renders the container ID on Docker Compose setups.
#283 — Duplicate server name in notification prefix and suffix suppressed.
#271 — Log sort order persists across navigation.
#265 — Stale update detection suppressed from pre-clear watcher scans.
#323 — Popovers on the Containers list rendered off-screen for the last row + drifted on scroll — Added a buildPopoverStyle helper that measures available space and flips the popover upward when needed; added a global scroll listener that closes both popovers.
Watcher "Last Run" display — Watchers page now shows relative timestamps for last run. (#189)
#187 — Agent column picker positioning fixed.
#184 — Dashboard confirm dialog — ConfirmDialog moved to global app shell so update prompts from the dashboard appear immediately.
Stack/group view no longer collapses to ungrouped mid-update when containers are recreated. loadGroups() now indexes the map under id, name, AND displayName.
#340 — Self-update no longer preserves stale Drydock version metadata. The self-update clone path now drops image-inherited environment variables and labels from the old image when the target image changed them.
#345 — Host names with numeric suffixes no longer lose the differentiating character in the Containers table.
Truncated release notes body now marked with trailing ellipsis (commit 3a9bd098).
Accepted update dispatch failures now logged (commit 674a0ed8).
Row update overlays anchored to first data cell width (commit 4bdb8d65).
Update state lost on navigation/refetch — Backend list endpoint now enriches containers with in-progress update operation state.
Digest-only image visibility — Watchers no longer silently drop containers with digest-only image references. Digest watch now stays enabled when Docker summary exposes a sha256:… image but container inspect recovers a tagged reference.
Same-name container update holds isolated to the correct instance (commit 02433a02). Added an identityKey discriminator so the hold follows the container's stable identity through id changes.
Security view container chooser traps keyboard focus (commit e98603c1). Added standard focus-trap: focus is moved to the first focusable element on open, Tab/Shift+Tab cycle within the popover, Escape returns focus to the previously-focused element.
Legacy xlink:href SVG attributes stripped by icon sanitizer (commit 0309bacb). href is now in the allowlist; the deprecated xlink:href form remains blocked.
Floating semver aliases excluded from greater-than check (commit 0b9eaaf3). isGreaterCandidateTag now requires strictly greater semver in one direction and not-greater in the reverse, so floating aliases like 3.3 and 3.3.0 drop out of the candidate set entirely.

Security

Mau registry auth scope is now percent-encoded. Mau.authenticate interpolated image.name into the JWT auth scope query parameter without encoding, diverging from the parent Gitlab provider which encodes it; a URL-significant character in an image name could corrupt the query or inject extra parameters. The scope is now encoded identically to Gitlab.

The following entries are hardening from a 2026-06-01 multi-agent security review (no critical/high findings).

OIDC UserInfo response is bound to the id_token subject on login (commit 661c21b9). The authorization-code login path called fetchUserInfo with skipSubjectCheck, omitting the OIDC Core 5.3.2 check that the UserInfo sub matches the token sub. The validated id_token subject is now enforced; the bearer-token path (no id_token, no reliable expected subject) is unchanged.
CSRF same-origin check no longer trusts forged X-Forwarded-Host (commit a132318e). getExpectedOrigin read X-Forwarded-Host/-Proto unconditionally, so a client could forge them to satisfy the mutation same-origin check even with trust proxy disabled. The forwarded host is now honored only when Express trust proxy is enabled, the protocol derives from the already-gated req.protocol, and the Host port is preserved for non-standard-port deployments. Upgrade note: this aligns enforcement with the long-documented requirement to set DD_SERVER_TRUSTPROXY behind a TLS-terminating reverse proxy (reverse-proxy setup). A deployment that terminated TLS at a proxy but never set DD_SERVER_TRUSTPROXY previously worked only because the unconditional header trust masked the misconfiguration; it must now set DD_SERVER_TRUSTPROXY (hop count, e.g. 1) or state-changing requests will return 403.
GitHub release-notes token withheld for untrusted source repos (commit 7186195c). A dd.source.repo container label (settable by anyone who controls the container) could redirect the operator's release-notes token — including the GHCR PAT fallback — to an arbitrary GitHub repo. Source-repo resolution now carries provenance, and no token is attached when the repo originates from that per-deployment label or from a persisted container.sourceRepo fallback without provenance; image-label, OCI-label, GHCR-path, and live Docker Hub metadata lookups remain trusted. A follow-up (commit 0a014304) also segregates the release-notes cache by trust (#auth/#anon), so an untrusted not-found result can no longer suppress a later trusted, token-bearing fetch for the same repo.
Playwright E2E image pinned by digest (commit 8234ef33). mcr.microsoft.com/playwright:v1.60.0-noble is now pinned by its sha256 digest so a mutated/republished tag cannot silently change the image the E2E job pulls and runs.
Container-query filter values constrained to primitives (commit a5ae7a89). Express's qs parser turns ?key[$regex]=… into a nested object, and the store's query-sanitizer only guarded prototype-pollution keys — so an authenticated container-list request could inject a LokiJS $regex (native RegExp → ReDoS, bypassing the project's re2js immunity) or $ne/$gt operators. Exact-match filter values are now restricted to primitives at the store choke point, neutralizing operator injection for every caller.
Security digest templates no longer evaluated as JavaScript (commit e74fc56d). renderSecurityDigestTemplate rendered the operator-supplied SECURITYDIGESTTITLE / SECURITYDIGESTBODY via new Function('scan', …), executing them as a template literal — arbitrary code execution in the digest path. The renderer now routes through the same sandboxed ${…} interpolation engine as every other trigger template; the per-severity lists and plural noun the default template relied on are pre-computed in code and exposed as scan.criticalList / scan.highList / scan.containerNoun, so default output is unchanged. The digest feature shipped only in the v1.5.0 RC train, so this is a within-cycle hardening; any RC user who set a custom digest template using JS expressions (arrow functions, array methods, arithmetic) should switch to the documented ${scan.*} variables, as unsupported expressions now resolve to an empty string instead of running.
Template method-call arguments split on top-level commas only (commit c485f412). safeInterpolate split method arguments on every comma, so a comma inside a quoted string literal or nested parentheses (e.g. ${x.replace(',', ';')}) was mis-parsed; argument splitting now respects quotes and parentheses, hardening the template parser the digest renderer depends on.
Registry token-fetch requests now honor operator TLS settings (commit dfbbd159). GAR, GitLab, Mau, DHI and public-ECR built their own token-fetch request and called axios() directly, bypassing withTlsRequestOptions() — so the credential exchange ignored the configured cafile / insecure / client-cert and validated against the system trust store. The shared TLS helper is now applied to those token fetches (Ecr was realigned to extend BaseRegistry).
Bearer token-endpoint requests now set maxRedirects: 0 to prevent credential exfiltration via HTTP redirect. A host-validated token endpoint returning a 3xx could otherwise redirect the operator's Basic credentials to an attacker-controlled host; setting maxRedirects: 0 causes axios/follow-redirects to fail closed on any redirect response, and the existing failClosedAuth catch path turns that into a clean auth failure without forwarding credentials. Applied to BaseRegistry.authenticateBearerFromAuthUrl and to the four providers (GAR, GitLab, Mau, DHI) that build their own credentialed token-fetch requests outside the shared helper. The malformed-realm fallback debug log is also sanitized to avoid logging attacker-controlled realm values verbatim (committed separately as 786c2317).
CSRF same-origin enforcement extended to authenticated /auth mutations (commit d1611f88). The /auth router sat outside the middleware chain that runs requireSameOriginForMutations, leaving bodyless POST /auth/logout and POST /auth/remember without same-origin protection (exploitable under DD_SERVER_COOKIE_SAMESITE=none).
OIDC username falls back to the sub claim before unknown (commit 8e00dfd3). Identities without an email claim previously collapsed to a single unknown username sharing one session-eviction bucket.
schemaVersion-1 manifest parsing guarded (commit fee0bbcc). A malformed/missing v1Compatibility from a registry threw an unhandled error that silently dropped the container from the watch cycle; it is now fully optional-chained and wrapped with a descriptive error.
Diagnostic debug dump redacts plural *_TOKENS_* env vars (commit 238727f5). The redaction set matched token but not tokens, so DD_SERVER_WEBHOOK_TOKENS_* values printed in plaintext.
Agent HTTP server is rate-limited before authentication (commit 8cddaeab). The shared-secret-gated agent endpoints had no throttle on repeated failed attempts; a 60s/300 limiter now sits ahead of the auth middleware (/health exempt; the long-lived SSE stream counts once on open).
Basic-auth string comparison no longer leaks length via timing (commit 6ff55463). timingSafeEqualString early-exited on length mismatch; both operands are now hashed to a fixed-length sha256 digest before timingSafeEqual, matching verifyShaPassword.
Quay pagination tokens percent-encoded (commit a19e0102). next_page/last values parsed from the registry Link header were appended to the request URL unencoded, allowing query-parameter injection from a malicious/buggy registry response.
Startup warning when falling back to a store-persisted session secret (commit 9431c1d2). When DD_SESSION_SECRET is unset, a one-time warning now recommends setting it explicitly and keeping the store directory non-world-readable.
Startup warning when trust proxy is set to boolean true (commit 71eac008). DD_SERVER_TRUSTPROXY=true trusts all X-Forwarded-For hops, letting clients spoof req.ip and evade per-IP login lockout; the warning steers operators to a hop count (=1).
TCP Docker host is validated before the self-update controller passes it to Dockerode (commit 441b4358). validateTcpDockerHost rejects values that contain a URL scheme prefix, a userinfo segment, whitespace, or path separators, throwing a descriptive error before any network connection is attempted.
Proxied SVG icons sanitized before caching (commit 54d93a3b). SVG payloads fetched from upstream icon CDNs are run through an allowlist-based sanitizer before being written to the icon cache.
Command action trigger env values sanitized to strip shell metacharacters (commit 1113d8ca). Container-derived values injected into the command subprocess environment are now stripped of shell metacharacters ($, `, ;, (, )) before the environment map is passed to execFile.
Credential redaction expanded to x-registry-auth, *-token, and api-key fields (commit 4417ce25). A second regex pass now redacts x-registry-auth, any field matching *-token, and api-key / api_key values before the payload leaves the server.
Credential status pattern matching uses RE2 (commit df9b914a). BaseRegistry.getRejectedCredentialStatus now uses RE2JS.compile(…) to maintain the project-wide ReDoS immunity guarantee.
Registry instances using insecure=true now log a warning on every request (commit cd14e3a9).
DD_SESSION_SECRET is now required; auto-generated and persisted on first boot when unset. (See Changed entry above for the full history of rc.20 vs rc.21.)
Agent connections over plain HTTP with a configured secret are now rejected at startup (commit 7c6f6c20). Use HTTPS or a TLS-terminating proxy for agent connections that require a shared secret.
GHCR token fallback treats whitespace-only tokens as missing (commit 711d583c). All token checks now call .trim().length > 0.
GHSA-xq3m-2v4x-88gg (critical) — Bumped protobufjs 7.5.4 → 7.5.5 to close the prototype-chain arbitrary-code-execution advisory.
GHSA-r4q5-vmmm-2653 (medium) — Bumped follow-redirects to 1.16.0 to close the custom-auth-header cross-domain leak advisory.
Hook command grammar validator — User-supplied pre-update / post-update hook commands are now validated against a restricted shell-safe grammar at config time.
OIDC authorization endpoint strict match — The OIDC flow now requires an exact match against the discovered authorization endpoint.
OIDC token redaction in error logs — Error log lines from the OIDC pipeline redact bearer / id / refresh tokens.
Rate-limit key derivation — Unauthenticated rate-limit buckets now key on socket.remoteAddress in preference to request.ip, eliminating the X-Forwarded-For spoof-ability.
CORS origin required when enabled — Enabling CORS now requires DD_SERVER_CORS_ORIGIN to be set explicitly.
OIDC redirect target allowlist — Post-login redirects are now validated against the backend's endpoint allowlist.
Healthcheck HTTPS probe hardening — /bin/healthcheck no longer uses popen() with shell command interpolation. The probe now locates the openssl binary explicitly, fork/execs it with pipes, and uses poll-driven I/O with SIGPIPE handling.
SSE log IP hashing with opt-in raw mode — SSE connect/disconnect lines log a salted hash of the source IP (h:xxxxxxxx) by default. Set DD_SSE_DEBUG_LOG_IP=true to temporarily log raw IPs.
HTTP trigger proxy URL scheme validation — HTTP trigger proxy URLs must now be http:// or https:// schemes.
Vulnerability CSV export escape hardening — Every CSV field is now quoted unconditionally, and tab/CR leading characters are escaped alongside =+-@.
Vite CVE patches — Updated vite to 8.0.7 (ui) and 7.3.2 (demo) to fix CVE-2026-39363, CVE-2026-39364, CVE-2026-39365.
Axios CVE-2025-62718 — Updated axios 1.13.6 → 1.15.0.
fast-xml-parser override 5.5.8 → 5.7.1 — Addresses GHSA-gh4j-gqv2-49f6 / CVE-2026-41650.
uuid 13.0.0 → 14.0.0 — Addresses GHSA-w5hq-g745-h8pq.
fast-xml-parser upgraded to 5.5.8 — Addresses CVE-2026-33349 (numeric entity expansion bypass).
Log injection prevention — Removed version string interpolation from startup and migration log messages.
Reflected XSS in Podman redirect guard — 404 handler no longer reflects request URL in response body.
WebSocket origin and lockout hardening — Added stricter WebSocket origin validation and safer lockout file-permission handling.
Agent log entry sanitization — Agent log proxy endpoint now uses an allowlist-based normalizer that only forwards known fields.
Security bouncer enforcement on container updates — Update and Update All actions now enforce the security bouncer gate.
Vulnerability URL and CSV sanitization — Vulnerability URLs validated before rendering; CSV export fields sanitized against formula injection.
Snyk policy file — Added a repo-level .snyk file for reviewed false-positive Snyk Code findings.
Supply-chain toolchain refresh — Bumped pinned Alpine edge/testing package versions for cosign and trivy in the Dockerfile.
Binary indices and drain concurrency cap for notification outbox (commit 9393253e). findReadyForDelivery fields switched to binary indices for O(log n) B-tree lookups. OutboxWorker gains a maxDrainConcurrency option (default 10) backed by a DrainSemaphore.

Dependencies

Vite 7.3 upgraded to 8.0 — Migrated to Vite 8.0 with Rolldown bundler.
Patch/minor dependency bumps — Updated all patch/minor dependencies and upgraded knip to v6.
Vulnerable transitive dependency patches — nodemailer 8.0.3→8.0.4, picomatch→4.0.4, brace-expansion→5.0.5, smol-toml→1.6.1, yaml→2.8.3, next 16.2.1→16.2.2 (CVE-2025-59472), lodash 4.17.23→4.18.1 (CVE-2026-2950, CVE-2026-4800).
Pinned pinia and vue-i18n to exact versions (commit fd0b02a5). Both were the only ^-ranged UI dependencies; pinned to the locked 3.0.4 / 11.4.2 to match the exact-pinning discipline used everywhere else.
re2js 1.2.3 → 2.8.3 (major). Upgraded the ReDoS-safe RE2 regex engine behind safeRegExp() (tag include/exclude/transform) across the 1.x → 2.x boundary. The only 2.0.0 breaking change — native-ECMAScript replaceAll/replaceFirst replacement semantics — is not exercised by drydock; the compiled compile/matcher/find/group surface is unchanged and Node ≥24 satisfies re2js 2.x's Node ≥18 floor. All 298 tag/regex tests pass unchanged.
Runtime/security dependency bumps — helmet 8.1.0→8.2.0, undici 8.2.0→8.3.0, ws 8.20.1→8.21.0, axios 1.16.0→1.16.1 (dependency + override), express-rate-limit 8.5.1→8.5.2, nodemailer 8.0.7→8.0.10, semver 7.8.0→7.8.1.

Documentation

v1.5.0 deprecation sweep. Migrated every documentation example and test fixture off the v1.5.0-deprecated DD_TRIGGER_* / dd.trigger.* prefixes onto canonical DD_NOTIFICATION_* + dd.notification.* (messaging providers) and DD_ACTION_* + dd.action.* (update executors). Touched 29 files in content/docs/current/**, the in-repo README roadmap, CONTRIBUTING, and all QA/CI/demo compose fixtures.
Guide/API endpoint synchronization — Updated current docs and guides to consistently use canonical /api/v1/* paths, expanded container list API docs, and added dashboard customization guide.
#342 follow-up — Registry env-var naming convention now explained in the registries index. A new "Naming registry instances" callout explains that the {REGISTRY_NAME} placeholder is a user-chosen label that namespaces multiple instances of the same registry type.
#342 follow-up — Watcher cron callout explains rate-limit interaction with hourly polling.

1.5.0-rc.29 — 2026-05-30

Fixed

Docker Compose update failure safety (#391, #407). Prevented a failed Compose update from destroying the running container.
ZAP SARIF path handling (#404, #405). Relativized HTTP artifact URIs through SARIF originalUriBaseIds and preserved the leading slash so root-path findings validate in GitHub Code Scanning.
DAST authentication alignment (#403). Updated the baseline DAST credentials to match the QA compose password hash.
Authentication readiness gate (#402). Closed a strategy-registration timing gap so /health reports readiness only after authentication strategies are registered.

1.4.5 — 2026-03-17

Added

Dashboard Update buttons — Per-row update buttons and "Update all" button in the Updates Available dashboard widget. (#173)
Getting Started guide — New step-by-step onboarding guide covering watchers, tag filters, registries, notifications, auto-updates, safety features, and multi-host setup. (#153)

Fixed

Container recreate alias filtering — Hardened Docker watcher timestamp parsing, added event handler early return for transient aliases, canonical MQTT topic naming, and stale topic cleanup for recreated containers. (#156)
About modal version display — Version is now fetched dynamically from the API instead of being hardcoded, ensuring the modal always reflects the running server version. (#167)
Version resolution fallback — DD_VERSION=unknown is now skipped so the version is correctly read from package.json at startup.
Theme circle transition origin — The theme toggle circle animation now originates from the click point instead of the viewport center.
Trigger code bugs — Gotify URL and Apprise URL now correctly enforce .required() validation; Kafka clientId casing normalized with clientId kept as a deprecated compatibility alias until v1.6.0.

Security

OIDC debug log redaction — Sensitive OIDC parameters (client_id, code_challenge, state, etc.) are now redacted in debug logs. (#168)
Agent API error sanitization — Error logs and responses in the agent API are sanitized to prevent leaking internal details.
Registry config value redaction — Trigger group configuration now logs keys only, not values, to prevent secret leakage.

Changed

API versioning — All UI fetch calls migrated from /api/ to /api/v1/ paths.
OIDC empty bearer token log level — Downgraded empty bearer token log from warn to debug. (#169)

Documentation

Docs audit (78 files) — Fixed 18 doc accuracy issues, 3 code bugs, 22+ broken links, and restructured 8 pages (90 callouts → 36). Triggers overview reduced from 17 callouts to 3 with threshold reference table. Template variables expanded from 12 to 30 with example values. Docker Compose trigger now linked from triggers overview. (#153, #172)

Dependencies

Security — fast-xml-parser 5.3.8 → 5.5.6 (CVE: numeric entity expansion bypass), next 16.1.6 → 16.1.7 (HTTP smuggling, CSRF bypass, DoS)
CI — step-security/harden-runner v2.15.1 → v2.16.0, github/codeql-action v4.32.6 → v4.33.0
App — @aws-sdk/client-ecr, @slack/web-api, express-rate-limit, fast-check, knip, nodemailer, undici, @types/node
UI — @iconify-json/lucide, knip, jsdom 28 → 29
Website — fumadocs-core, fumadocs-mdx, fumadocs-ui, lucide-react, lefthook, @vercel/analytics v1 → v2, @vercel/speed-insights v1 → v2
Demo — @iconify-json/lucide, @vitejs/plugin-vue, msw 2.10 → 2.12
E2E — @dotenvx/dotenvx, fast-xml-parser, minimatch

Chore

QA infrastructure — Added Portainer, slow-shutdown container, and watchevents service for end-to-end container recreate testing. (#156)
Coverage and types cleanup — Coverage read retry, OpenAPI re-export, and auth types cleanup.
Demo version bump — Bumped demo package version to 1.4.5.
Alias filtering test coverage — Added timestamp edge-case tests for Docker watcher alias filtering. (#156)
msw worker regenerated — Updated mockServiceWorker.js for msw 2.12 (origin-based → clientId-based validation).

1.4.4 — 2026-03-16

Added

Click-to-copy on version tags — CopyableTag component with clipboard feedback on all version displays (dashboard, container list, detail panels). (#164)
Dark mode icon inversion — Simple Icons (si: prefix) auto-invert in dark mode via Tailwind dark variant.
Tailwind v4 class-based dark mode — @custom-variant dark directive for proper .dark class detection.

Changed

Dashboard Updates Available version column centered.
Sidebar search button border removed — ⌘K badge improved dark mode contrast.
URL rebrand — All URLs updated from drydock.codeswhat.com to getdrydock.com.

Fixed

Dashboard host-status widget showing 0 for non-agent remote watchers — Dashboard host-status widget incorrectly showed zero container counts for non-agent remote watchers by using watcher.id instead of watcher.name as the lookup key. (#155)
Container recreate alias duplicates — Unconditional 30s transient window skip, single inspect per event, event path guard prevent Docker's transient <id-prefix>_<name> aliases from producing duplicate container entries in triggers. (#156)
Tooltip viewport overflow — Replaced CSS pseudo-element tooltips with body-appended popup using position:fixed and auto-flip. (#165)
Theme switcher broken — Restored document binding for startViewTransition API.

1.4.3 — 2026-03-16

Fixed

DNS resolution failures on Alpine (EAI_AGAIN) — Node.js 24 defaults to verbatim DNS ordering, which on Alpine's musl libc can cause getaddrinfo EAI_AGAIN errors when IPv6 records are returned first on dual-stack networks. Drydock now defaults to IPv4-first DNS ordering at startup, configurable via DD_DNS_MODE (ipv4first | ipv6first | verbatim, default: ipv4first). (#161)
Multi-platform Docker build validation — CI now validates multi-platform Docker builds to catch architecture-specific failures early.

Security

Zizmor findings blocking in CI and lefthook — GitHub Actions security findings from zizmor are now blocking in both CI and local pre-push hooks.
Secrets scoped to dedicated GitHub environments — CI secrets are now scoped to dedicated GitHub environments instead of repository-level access.

Documentation

Docker socket security guide — Expanded watcher docs with comprehensive security section: comparison table of all access methods, socket proxy setup (recommended), remote Docker over TLS with cert generation walkthrough, rootless Docker guide, full Docker API endpoint reference showing exactly which endpoints Drydock uses for read-only monitoring vs write operations (updates).

Dependencies

CI actions — Bumped upload-artifact to v7 and replaced nick-fields/retry.

1.4.2 — 2026-03-15

Added

Update-operations pagination — GET /api/containers/:id/update-operations now supports limit and offset query parameters with _links navigation, matching the existing container list pagination pattern.
Periodic audit log pruning — Audit store now runs a background timer (hourly, unref'd) to prune stale entries even with low insert volume, in addition to the existing insert-count-based pruning.
Container release notes enrichment — Watch cycles now enrich update candidates with GitHub release metadata (result.releaseNotes and result.releaseLink), and full notes are available via GET /api/containers/:id/release-notes.
Container list sort and filter query params — GET /api/containers now supports sort (name, status, age, created, with optional - prefix for descending), plus status, kind, watcher, and maturity filters.
Update age and maturity API signals — Container payloads now track updateAge and support maturity states (hot, mature, established) for filtering and policy workflows.
Suggested semver tag hints — Containers tracked on non-semver tags such as latest now expose result.suggestedTag to surface the best semver target.
oninclude trigger auto mode — Trigger auto now supports oninclude, which only auto-runs for containers explicitly matched by include labels. (#160)
Container runtime observability APIs — Added /api/containers/stats, /api/containers/:id/stats, and /api/containers/:id/stats/stream for resource telemetry, plus richer per-container runtime snapshots.
Real-time container log streaming — Added WebSocket streaming at /api/v1/containers/:id/logs/stream with stdout/stderr, tail, since, and follow controls.
Container logs and runtime stats panels — Container detail views now include dedicated live Logs and Runtime Stats tabs (including full-page logs route support).
Signed registry webhook receiver — Added POST /api/webhooks/registry with HMAC signature verification and provider-specific payload parsing for registry push events.
Auth lockout Prometheus observability — Added Prometheus metrics for login success/failure and lockout activity.

Changed

Extract maturity-policy module — Consolidated scattered day-to-millisecond conversions and maturity policy constants into app/model/maturity-policy.ts, shared by backend, UI, and demo app.
Extract security-overview module — Moved security vulnerability aggregation and pagination logic from crud.ts into dedicated app/api/container/security-overview.ts for improved readability and testability.
Refactor Docker Compose trigger — Extracted YAML parsing/editing into ComposeFileParser and post-start hook execution into PostStartExecutor, reducing the monolithic Dockercompose.ts by ~400 lines.
Decompose useContainerActions — Split the 1200-line composable into focused modules: useContainerBackups, useContainerPolicy, useContainerPreview, and useContainerTriggers.
Registry error handling — Replaced catch (e: any) with catch (e: unknown) and getErrorMessage(e) in component registration and trigger/watcher startup.
E2E test resilience — Container row count assertions now use toBeGreaterThan(0) instead of hardcoded counts, preventing false failures when the QA environment has a different number of containers.
Registry digest cache dedup per poll cycle — Digest cache lookups now deduplicate repeated requests within a single poll cycle, reducing redundant registry calls and improving metric accuracy.
Extract runtime config evaluation context type — Consolidated scattered runtime field evaluation parameters into a typed ClonedRuntimeFieldEvaluationContext interface for trigger providers.
Argon2 hash parsing type safety — Extracted Argon2Parameters interface, parameter key type guard, and PHC parameter parsing into a reusable function for improved type safety.
Extract agent client initialization methods — Extracted URL parsing, HTTPS detection, protocol validation, and TLS configuration from monolithic constructor into focused private methods.
Extract shared self-hosted registry config schema — Deduplicated the registry configuration schema (url, login, password, auth, cafile, insecure, clientcert, clientkey) into a reusable helper shared by Custom and SelfHostedBasic registry providers.

Documentation

Podman docs expansion — Added Podman setup/compatibility guidance plus SELinux, podman-compose, and production notes in watcher and FAQ docs.
Docs site URL rebrand — Replaced drydock.codeswhat.com links with getdrydock.com across docs pages, sitemap/robots metadata, and website copy.

Fixed

CSRF validation behind reverse proxies — Same-origin mutation checks now honor X-Forwarded-Proto and X-Forwarded-Host when present before falling back to direct request protocol/host, preventing false 403 CSRF validation failed responses in TLS-terminating proxy setups. (#146)
Hosts page missing env-var-configured watchers — The Hosts page hardcoded a single "Local" entry and only added agent-based hosts. Watchers configured via DD_WATCHER_* environment variables (e.g. remote Docker hosts) were never displayed, even though their containers appeared correctly on the Containers page. The page now fetches all watchers from the API and displays each local watcher with its actual name and connection address. (#151)
Docker events reconnect with malformed errors — Reconnect failure logging no longer crashes when the error object has no message property.
Security overview container count — The security vulnerability overview now uses the lightweight store count instead of loading all container objects just to count them.
Compose path deduplication — Replaced indexOf-based dedup with Set for compose file path detection in container security view.
Build provenance attestation — CI attestation step now runs with if: always() so provenance is attested even when prior optional steps are skipped.
Container recreate alias duplicates in triggers — When containers were recreated externally (via Portainer or docker compose up), Docker's transient <id-prefix>_<name> aliases were treated as new containers, producing duplicate entries in MQTT Home Assistant discovery sensors and Telegram notifications. Added alias deduplication filtering and force-removal of stale container IDs on recreate detection. (#156)
Stale store data after external container recreation — Watch-at-start scan was suppressed when the store already had container data, leaving stale records from the previous run after external container recreation. Removed the suppression so every startup runs a full scan with alias filtering. (#157)
Watcher container counts on Hosts page — Per-watcher container counts on the Hosts page used watcher.id (e.g. docker.esk83) as the lookup key instead of watcher.name (e.g. esk83), causing counts to display as zero for env-var-configured watchers. (#155)
Docker images tagged :main with no version — CI release workflow triggered on both main branch pushes and version tags, producing Docker images tagged :main that showed DD_VERSION=main instead of a real version number. Release workflow now only triggers on version tags (v*). (#154)
Maturity badge sizing and tooltip clipping — Fixed maturity badge height mismatch with other text badges and removed overflow-hidden from DataListAccordion that clipped tooltips in list view.

Security

Agent secret over plaintext HTTP warning — Agent clients now log a warning when a shared secret is configured over unencrypted HTTP, advising HTTPS configuration.
Auth audit log injection — Login identity values are now sanitized with sanitizeLogParam() before inclusion in audit log details, preventing log injection via crafted usernames.
SSE self-update ack hardening — Added validation for empty clientId/clientToken, non-ack broadcast mode, and client-not-bound-to-operation rejection.
FAQ: removed insecure seccomp advice — Removed the "Core dumped on Raspberry PI" FAQ entry that recommended --security-opt seccomp=unconfined, which completely disables the kernel's syscall sandbox. The underlying libseccomp2 bug was fixed in all supported OS versions since 2021.

Dependencies

biome — Bumped to 2.4.7 with import ordering fixes for new lint rules.
vitest — Bumped to 4.1.0 in app workspace with fast-check/vitest 0.3.0 and knip 5.86.0.
UI packages — Bumped vue, vitest, storybook, and icon packages.
CI actions — Bumped zizmor-action to v0.5.2 and cosign-installer to v4.1.0.

1.4.1 — 2026-03-14

Added

Headless mode (DD_SERVER_UI_ENABLED) — Run drydock as an API-only service by setting DD_SERVER_UI_ENABLED=false. The REST API, SSE, and healthcheck endpoints remain fully functional while the UI is not served. Useful for controller nodes that only manage agents.
Maturity-based update policy — Per-container update maturity policy configurable from the container row's Update policy menu (or PATCH /api/v1/containers/:id/update-policy). Modes: all (default — allow any update) and mature (block updates detected less than maturityMinAgeDays ago, default 7). UI shows NEW/MATURE badges with flame/clock icons on containers with available updates. (#120)
?groupByStack=true URL parameter — Bookmarkable URL parameter to enable stack grouping on the containers page. Also accepts ?groupByStack=1. (#145)

Changed

Connection-lost overlay animation — The connection-lost and reconnecting overlays now use a bounce animation for improved visual feedback.
Watch target resolution refactored to discriminated union — Internal refactor of the watch target resolution logic for improved type safety and maintainability.

Fixed

Agent handshake and SSE validation failure — Fixed agent API returning redacted container data (with sensitive field on env entries) instead of raw data, causing controller-side Joi validation to reject the handshake and crash on real-time SSE container events. SSE payloads now prefer canonical raw store data with sanitization fallback. (#141)
Mangled argon2 hash detection — Docker Compose $ interpolation can strip $ delimiters from argon2 PHC hashes, producing an invalid hash that silently failed registration. Drydock now detects mangled hashes at startup and surfaces an actionable error message. (#147)
Anonymous auth fallback — When all configured auth providers fail to register (e.g. due to a mangled hash), Drydock now falls back to anonymous mode if DD_ANONYMOUS_AUTH_CONFIRM=true is set, instead of leaving the user with no way to log in. (#147)
Auth registration errors on login page — Registration warnings (e.g. invalid hash format) are now surfaced on the login page so users can see exactly what went wrong instead of a generic "No authentication methods configured" message. (#147)
CSP inline style violations — Replaced runtime element.style mutations (DataTable column resize, tooltip directive, theme transitions, preference restore) with CSS custom properties and class-based styling. Relaxed style-src to include 'unsafe-inline' for vendor libraries (iconify-icon, Vue Transition) that set element.style programmatically.
Compose trigger affinity across remapped roots — Compose trigger file-path matching now uses suffix-based comparison as a fallback when the container's compose label path (host mount) differs from the trigger's configured path (container-internal), preventing missed trigger associations in bind-mount setups.
Compose trigger association — Enforce compose-file affinity when associating triggers with containers, preventing incorrect trigger-container matching. (#139)
Update All button icon centering — Fixed icon-text alignment in the Update All group header button to match the split button pattern used elsewhere.
Selected card border clipping — Fixed card border clipping on the first grid column in card view.

Security

Username enumeration timing side-channel — Eliminated timing difference between valid and invalid usernames during authentication.
LokiJS metadata exposure — Stripped internal LokiJS fields ($loki, meta) from settings API responses, agent API container responses, and /app endpoint.
Permissions-Policy header — Added Permissions-Policy header to restrict browser feature access (camera, microphone, geolocation, etc.).
CSP and Cross-Origin-Embedder-Policy — Tightened Content Security Policy and added COEP header.
Production image hardening — Removed wget, nc, and npm from the production Docker image; upgraded zlib.

Dependencies

undici — Bumped to 7.24.1 (fixes 12 CVEs including WebSocket memory consumption, CRLF injection, and request smuggling).

1.4.0 — 2026-02-28

Breaking Changes

MQTT HASS_ATTRIBUTES default changed from full to short — This changes Home Assistant entity payloads by default, excluding large SBOM documents, scan vulnerabilities, details, and labels. To retain the previous payload behavior, set DD_TRIGGER_MQTT_{name}_HASS_ATTRIBUTES=full explicitly.

Added

Audit log for container state changes — External container lifecycle events (start, stop, restart via Portainer or CLI) now generate container-update audit entries with the new status, so the audit log reflects all state changes, not just Drydock-initiated actions. (#120)
mTLS client certificate support — Registry providers now accept CLIENTCERT and CLIENTKEY options for mutual TLS authentication with private registries that require client certificates.

Backend / Core

Container recent-status API — GET /api/containers/recent-status returns pre-computed update status (updated/pending/failed) per container, replacing the client-side audit log scan and reducing dashboard fetch payload size.
Dual-slot security scanning — "Scan Now" automatically scans both the current running image and the available update image when an update exists. Results are stored in separate slots (scan/updateScan) and the Security page shows a delta comparison badge (+N fixed, -N new) next to each image that has both scans.
DD_LOG_BUFFER_ENABLED toggle — Disable the in-memory log ring buffer via DD_LOG_BUFFER_ENABLED=false to reduce per-log processing overhead. When disabled, /api/log/entries returns an empty array. Defaults to true.
Scheduled security scanning — Set DD_SECURITY_SCAN_CRON to automatically scan all watched containers on a cron schedule. DD_SECURITY_SCAN_JITTER (default 60s) spreads load with random delay before each cycle.
Security scheduler shutdown on exit — Security scan scheduler is now explicitly shut down during graceful exit, preventing orphan timers from delaying process termination.
On-demand sensitive env value reveal — Container environment variables are redacted by default in API responses. Individual values can be revealed on-demand via /api/containers/:id/env/reveal with audit logging.
On-demand scans populate digest cache — Manual container scans now populate the digest-based dedup cache, preventing redundant rescans of the same image digest.
Per-container webhook opt-out — New dd.webhook.enabled=false container label to exclude individual containers from webhook triggers without disabling the webhook API globally.
Scan cancellation and mobile scan progress — Security batch scans can now be cancelled mid-flight. Mobile scan progress UI improved with responsive layout.
Security scan coverage counts — Security view header shows scanned/total container counts for at-a-glance scan coverage.
Notification rule management API and persistence — /api/notifications CRUD endpoints backed by LokiJS-persisted notification rules for update-available, update-applied, update-failed, security-alert, and agent-disconnect event types.
Rule-aware runtime dispatch — Trigger event dispatch resolves notification rules at runtime so per-event enable/disable and trigger assignments actively control which triggers fire.
Security-alert and agent-disconnect events — New event types with audit logging and configurable deduplication windows. Security alerts fire automatically on critical/high vulnerability scan results.
Compose-native container updates — Compose-managed containers now update via docker compose up -d lifecycle instead of Docker API recreate, preserving compose ownership and YAML formatting.
Rename-first rollback with health gates — Non-self container updates use a rename-first strategy (rename old → create new → health-gate → remove old) with crash-recoverable state persisted in a new update-operation store collection. Rollback telemetry via dd_trigger_rollback_total{type,name,outcome,reason} counter.
Tag-family aware semver selection — Docker watcher infers the current tag family (prefix/suffix/segment style) and keeps semver updates within that family by default, preventing cross-family downgrades like 5.1.4 → 20.04.1. Added dd.tag.family label (strict default, loose opt-out) and imgset support. (#104)
Entrypoint/cmd drift detection — Docker trigger detects whether entrypoint/cmd were inherited from the source image vs user-set, replacing inherited values with target image defaults during update. Adds dd.runtime.entrypoint.origin and dd.runtime.cmd.origin labels.
Self-update controller with SSE ack flow — Dedicated controller container for self-update replaces the shell helper pattern. UI acknowledgment via SSE with operation ID tracking.
Server-issued SSE client identity — Replaced client-generated UUIDs with server-issued clientId/clientToken pairs for self-update ack validation, preventing spoofed acknowledgments.
config migrate CLI — node dist/index.js config migrate converts legacy WUD_* and Watchtower env vars/labels to DD_*/dd.* format across .env and compose files. Supports --dry-run preview and --source / --file selection.
Legacy compatibility usage metric — Prometheus counter dd_legacy_input_total{source,key} tracks local runtime consumption of legacy inputs (WUD_* env vars, wud.* labels) without external telemetry. Startup warns when legacy env vars are detected; watcher/trigger paths emit one-time deprecation warnings on wud.* label fallback.
Bundled selfhst icons for offline startup — Common container icons (Docker, Grafana, Nextcloud, etc.) bundled in the image so the UI works without internet on first boot.
Runtime tool status endpoint — /api/server/security/runtime reports Trivy/Cosign availability for the Security view.
Gzip response compression — Configurable via DD_SERVER_COMPRESSION_ENABLED and DD_SERVER_COMPRESSION_THRESHOLD (default 1024 bytes), with automatic SSE exclusion.
Container runtime details — Ports, volumes, and environment exposed in the container model and API for the detail panel.
Update detected timestamp — updateDetectedAt field tracks when an update was first seen, preserved across refresh cycles.
No-update reason tracking — result.noUpdateReason field surfaces why tag-family or semver filtering suppressed an available update.
Remove individual skip entries — remove-skip policy action allows removing a single skipped tag or digest without clearing all skips.
Update-operation history API — GET /api/containers/:id/update-operations returns persisted update/rollback history for a container.
Settings backend — /api/settings endpoints with LokiJS collection for persistent UI preferences (internetless mode). Icon proxy cache with atomic file writes and manual cache clear.
SSE real-time updates — Server-Sent Events push container state changes to the UI without polling.
Remember-me authentication — Persistent login sessions via remember-me checkbox on the login form.
Docker Compose trigger — Refresh compose services via Docker Compose CLI when updates are detected.
Advisory-only security scanning — DD_SECURITY_BLOCK_SEVERITY=NONE runs vulnerability scans without blocking updates. Scan results remain visible in the Security view and audit log.
OpenAPI 3.1.0 specification and endpoint — Machine-readable API documentation originally shipped on the unversioned OpenAPI route, covering all v1.4 endpoints with request/response schemas. Current releases expose the canonical spec at GET /api/v1/openapi.json.
Watcher agent support initialization — Watchers now initialize agent support on startup for distributed monitoring readiness.
Security vulnerability overview endpoint — New GET /api/containers/security/vulnerabilities returns pre-aggregated vulnerability data grouped by image with severity summaries, so the Security view no longer needs to load all containers.
MQTT attribute filtering for Home Assistant — MQTT trigger supports attribute-based filtering for Home Assistant integration, allowing selective publishing based on container attributes.
Docker Compose post_start env validation — Docker Compose trigger validates environment variables in post_start hooks before execution, preventing runtime errors from missing or invalid env var references.
MQTT HASS entity_picture from container icons — When Home Assistant HASS discovery is enabled, entity_picture is now automatically resolved from the container's dd.display.icon label. Icons with sh:, hl:, or si: prefixes map to jsDelivr CDN URLs for selfhst, homarr-labs, and simple-icons respectively. Direct HTTP/HTTPS URLs pass through unchanged. (#138)
dd.display.picture container label — New label to override the MQTT HASS entity_picture URL directly. Takes precedence over icon-derived pictures when set to an HTTP/HTTPS URL.

UI / Dashboard

Tailwind CSS 4 UI stack — Complete frontend migration from Vuetify 3 to Tailwind CSS 4 with custom shared components. All 13 views rebuilt with Composition API.
Shared data components — Reusable DataTable, DataCardGrid, DataListAccordion, DataFilterBar, DetailPanel, DataViewLayout, and EmptyState components used consistently across all views with table/cards/list view modes.
6 color themes — One Dark (clean/balanced), GitHub (clean/familiar), Dracula (bold purple), Catppuccin (warm pastels), Gruvbox (retro earthy warmth), and Ayu (soft golden tones). Each with dark and light variants. Circle-reveal transition animation between themes.
7 icon libraries — Phosphor Duotone (default), Phosphor, Lucide, Tabler, Heroicons, Iconoir, and Font Awesome. Switchable in Config > Appearance with icon size slider.
6 font families — IBM Plex Mono (default/bundled), JetBrains Mono, Source Code Pro, Inconsolata, Commit Mono, and Comic Mono. Lazy-loaded from local /fonts/ directory with internetless fallback.
Command palette — Global Cmd/Ctrl+K search with scope filtering (/ pages, @ runtime, # config), keyboard navigation, grouped sections, and recent history.
Notification rules management view — View, toggle, and assign triggers to notification rules with direct save through /api/notifications.
Audit history view — Paginated audit log with filtering by container, event text, and action type. Includes security-alert and agent-disconnect event type icons.
Container grouping by stack — Collapsible sections grouping containers by compose stack with count and update badges.
Container actions tab — Detail panel tab with update preview, trigger list, backup/rollback management, and update policy controls (skip tags, skip digests, snooze).
Container delete action — Remove a container from tracking via table row or detail panel.
Container ghost state during updates — When a container is updated, stopped, or restarted, its position is held in the UI with a spinner overlay while polling for the recreated container, preventing the "disappearing container" UX issue. (#80)
Skip update action — Containers with pending updates can be individually skipped, hiding the update badge for the current session without requiring a backend endpoint.
Slide-in detail panels on all views — Row-click detail panels for Watchers, Auth, Triggers, Registries, Agents, and Security views.
Interactive column resizing — Drag-to-resize column handles on all DataTable instances.
Dashboard live data and drag-reorder — Stat cards (containers, updates, security, registries) computed from real container data with drag-reorderable layout and localStorage persistence. Security donut chart, host status, and update breakdown widgets.
Log viewer auto-fetch and scroll lock — Configurable auto-fetch intervals (2s/5s/10s/30s) with scroll lock detection and resume for both ConfigView logs and container logs.
Keyboard shortcuts — Enter/Escape for confirm dialogs, Escape to close detail panels.
SSE connectivity overlay — Connection-lost overlay with self-update awareness and auto-recovery.
Login page connectivity monitor — Polls server availability and shows connection status on the login screen.
Server name badge for remote watchers — Shows the watcher name instead of "Local" for multi-host setups.
Dynamic dashboard stat colors — Color-coded update and security stats based on severity ratio.
About Drydock modal — Version info and links accessible from sidebar.
View wiring — Watcher container counts, trigger Test buttons with success/failure feedback, host images count, and registry self-hosted port matching all wired to live API data.
Font size preference — Adjustable font size slider in Config > Appearance for UI-wide text scaling.
Announcement banner — Dismissible banner component for surfacing release notes and important notices in the dashboard.
Dashboard vulnerability sort by severity — Top-5 vulnerability list on the dashboard now sorted by total count descending with critical count as tiebreaker, so the most severe containers appear first.
Rollback confirmation dialog — Container rollback actions now require explicit confirmation through a danger-severity dialog before restoring from backup.
Update confirmation dialog — Container update actions now require explicit confirmation through a dialog before triggering an update.
SHA-1 hash deprecation banner — Dashboard shows a dismissible deprecation banner when legacy SHA-1 password hashes are detected, prompting migration to argon2id.
Config tab URL deep-linking — Config view tab selection syncs to the URL query parameter, enabling shareable direct links to specific config tabs.

Changed

Compose trigger uses Docker Engine API — Compose-managed container updates now use the Docker Engine API directly (pull, stop, recreate) instead of shelling out to docker compose / docker-compose CLI. Eliminates spawn docker ENOENT errors in environments without Docker CLI binaries installed.
Compose self-update delegates to parent orchestrator — Self-update for compose-managed Drydock containers now uses the parent Docker trigger's helper-container transition with health gates and rollback, instead of direct stop/recreate.
Compose runtime refresh extracted — Shared refreshComposeServiceWithDockerApi() helper eliminates the recursive recreateContainer → updateContainerWithCompose call chain. Both code paths now converge on the same explicit, non-recursive method.
Compose-file-once batch mode re-enabled — COMPOSEFILEONCE=true now works with the Docker Engine API runtime. First container per service gets a full runtime refresh; subsequent containers sharing the same service skip the refresh.
Self-update controller testable entrypoint — Extracted process-level entry logic to a separate entrypoint module, making the controller independently testable without triggering process-exit side effects.
Dockercompose YAML patching simplified — Removed redundant type guards and dead-code branches from compose file patching helpers, reducing code paths and improving maintainability.
Dashboard fetches recent-status from backend — Dashboard now fetches pre-computed container statuses from /api/containers/recent-status instead of scanning the raw audit log client-side.
Prometheus collect() callback pattern — Switched container gauge from interval-based polling to the Prometheus collect() callback, letting Prometheus control collection timing and eliminating the background 5s timer.
Container security API refactored — Container security routes refactored into a dedicated module with type-safe SecurityGate integration, concurrent scan limiting (max 1), and trivy DB status-based cache invalidation.
DashboardView composable extraction — Extracted 700+ line monolith into useDashboardData, useDashboardComputed, useDashboardWidgetOrder, and shared dashboardTypes for better testability and separation of concerns.
Event-driven connectivity polling — AppLayout SSE connectivity monitoring now starts on disconnect and stops on reconnect instead of running a fixed interval, reducing unnecessary network requests.
Vulnerability loading optimized — Vulnerability data loaded from the container list API payload (includeVulnerabilities flag) instead of separate per-container fetches, reducing API calls on the Security view.
Default log format is JSON — Official Docker image now defaults to DD_LOG_FORMAT=json for structured production logs. Override with DD_LOG_FORMAT=text for pretty logs.
Scan endpoint rate limit reduced — POST /api/containers/:id/scan rate limit lowered from 100 to 30 requests/min to prevent resource exhaustion during aggressive scanning.
Single Docker image — Removed thin/heavy image variants; all images now bundle Trivy and Cosign.
Removed Vuetify dependency — All Vuetify imports, components, and archived test files removed. Zero Vuetify references remain.
Fail-closed auth enforcement — Registry bearer-token flows error on token endpoint failures instead of falling through to anonymous. HTTP trigger auth errors on unsupported types. Docker entrypoint requires explicit DD_RUN_AS_ROOT + DD_ALLOW_INSECURE_ROOT for root mode.
Fail-closed anonymous auth on fresh installs — New installs with no authentication configured and no DD_ANONYMOUS_AUTH_CONFIRM=true fail closed at startup (all API calls return 401). Users upgrading from a previous version are allowed anonymous access with a startup warning. Set DD_AUTH_BASIC_<name>_USER / DD_AUTH_BASIC_<name>_HASH to configure authentication, or set DD_ANONYMOUS_AUTH_CONFIRM=true to explicitly allow anonymous access.
Dashboard streamlined — Stat cards reduced from 7 to 4 (Containers, Updates, Security, Registries). Recent Activity widget removed to fit on single viewport. Background refresh prevents loading flicker on SSE events.
Notifications view is full rule management — Editable notification rules (enable/disable and trigger assignments) that save directly through /api/notifications.
Standardized API responses with collection pattern — All collection endpoints (/api/containers, /api/registries, /api/watchers, /api/authentications, /api/audit) now return { data: [...], total } instead of raw arrays. Supports pagination via offset/limit query parameters.
Paginated API discoverability links — Paginated collection responses now include _links.self and _links.next where applicable (for example /api/containers and /api/audit) to make page traversal explicit for API consumers.
Versioned API base path with transition alias — /api/v1/* is now the canonical API path for integrations. /api/* remains as a backward-compatible alias during migration and is planned for removal in a future major release (target: v2.0.0). [Editorial note: the removal target was later brought forward to v1.6.0 — see deprecations.]
Agent-scoped path order normalized — Agent-qualified component and trigger routes now use /:type/:name/:agent (for example /api/triggers/:type/:name/:agent and /api/containers/:id/triggers/:triggerType/:triggerName/:triggerAgent) instead of the old agent-first order. This is a breaking change for external API clients using the old path shape.
Machine-readable API error contract — All error responses return a consistent { "error": "..." } JSON structure with an optional details object for contextual metadata.
6 color themes — Replaced original Drydock theme with popular editor palettes: One Dark, GitHub, Dracula, Catppuccin, Gruvbox, and Ayu. Each with dark and light variants.
Argon2id password hashing — Basic auth now uses argon2id (OWASP recommended) via Node.js built-in crypto.argon2Sync() instead of scrypt for password hashing. Default parameters: 64 MiB memory, 3 passes, parallelism 4.
PUT /api/settings deprecated — PUT /api/settings now returns RFC 9745 Deprecation and RFC 8594 Sunset headers. Use PATCH /api/settings for partial updates. PUT alias removal targeted for v1.5.0. [Editorial note: the removal target was later moved to v1.6.0 — see deprecations.]
Basic auth argon2id PHC compatibility — Basic authentication now accepts PHC-format argon2id hashes ($argon2id$v=19$m=...,t=...,p=...$salt$hash) in addition to the existing Drydock argon2id$memory$passes$parallelism$salt$hash format. Hash-generation guidance now recommends the standard argon2 CLI command first, with Node.js as a secondary option.
Borderless UI redesign — Removed borders from all views, config tabs, detail panels, and shared data components for a cleaner visual appearance.
Dashboard version column alignment — Version column in the dashboard updates table is now left-aligned for better readability.
Detail panel expand button redesigned — Full-page expand button in the detail panel now uses a frame-corners icon instead of the previous maximize icon.
Sidebar active indicator removed — Removed the blue active indicator bar from sidebar navigation items for a cleaner look.

Fixed

Log level setting had no effect — DD_LOG_LEVEL=debug was correctly parsed but debug messages were silently dropped because pino's multistream destinations defaulted to info level. Stream destinations now inherit the configured log level. (#134)
Server feature flags not loaded after login — Feature flags (containeractions, delete) were permanently stuck as disabled when authentication was required, because the pre-login bootstrap fetch failure marked the flags as "loaded" and never retried. Now failed fetches allow automatic retry after login. (#120)
Compose trigger silently skips containers — Multiple failure paths in the compose trigger were logged at debug level, making it nearly impossible to diagnose why a trigger reports success but containers don't update. Key diagnostic messages (compose file mismatch, label inspect failure, no containers matched) promoted to warn level, and the "already up to date" message now includes container names. (#84)
Fallback icon cached permanently — The Docker placeholder icon was served with immutable cache headers, causing browsers to cache it permanently even after the real provider icon becomes available. Fallback responses now use no-store.
Basic auth upgrade compatibility restored — Basic auth now accepts legacy v1.3.9 Basic auth hashes ({SHA}, $apr1$ / $1$ , crypt, and plain fallback) to preserve smooth upgrades. Legacy formats remain deprecated and continue showing a migration banner, with removal still planned for v1.6.0.
Compose trigger rejects lowercase env var keys — Configuration keys like COMPOSEFILEONCE, DIGESTPINNING, and RECONCILIATIONMODE were lowercased by the env parser but the Joi schema expected camelCase. Schema now maps lowercase keys to their camelCase equivalents. (#120)
Compose trigger strips docker.io prefix — When a compose file uses an explicit docker.io/ registry prefix, compose mutations now preserve it instead of stripping it to a bare library path. (#120)
Compose trigger fails when FILE points to directory — DD_TRIGGER_DOCKERCOMPOSE_{name}_FILE now accepts directories, automatically probing for compose.yaml, compose.yml, docker-compose.yaml, or docker-compose.yml inside the directory. (#84)
Container healthcheck fails with TLS backend — The Dockerfile healthcheck now detects DD_SERVER_TLS_ENABLED=true and switches to curl --insecure https:// for self-signed certificates. Also skips the healthcheck entirely when DD_SERVER_ENABLED=false. (#120)
Agent CAFILE ignored without CERTFILE — The agent subsystem now loads the CA certificate from CAFILE even when CERTFILE is not provided, fixing TLS verification for agents behind reverse proxies with custom CA chains.
Service worker accepts cross-origin postMessage — The demo service worker now validates postMessage origins against the current host, preventing potential cross-origin message injection.
Action buttons disable and show spinner during in-progress actions — Container action buttons (Stop, Start, Restart, Update, Delete) now show a disabled state with a spinner while the action runs in the background, providing clear visual feedback. The confirm dialog closes immediately on accept instead of blocking the UI.
Command palette clears stale filter on navigation — Navigating to a container via Ctrl+K search now clears the active filterKind, preventing stale filter state from hiding the navigated container.
Manual update button works with compose triggers — The update container endpoint now searches for both docker and dockercompose trigger types, matching the existing preview endpoint behavior. Previously, users with only a compose trigger saw "No docker trigger found for this container".
CI: qlty retry on timeout — Changed retry_on from error to any so qlty timeouts trigger retries. Increased timeout from 5 to 8 minutes.
OIDC docs clarified DD_PUBLIC_URL requirement — OIDC documentation now explicitly marks DD_PUBLIC_URL as required and includes it in all provider example configurations (Authelia, Auth0, Authentik, Dex). Without this variable, the OIDC provider fails to register at startup.
Compose trigger docs updated for Engine API — Removed stale references to docker compose config --quiet CLI validation and docker compose pull/up commands. Docs now reflect in-process YAML validation and Docker Engine API runtime. Added callout that Docker Compose CLI is not required.
Container actions docs updated for compose triggers — Update action documentation now mentions both Docker and Docker Compose trigger support.
Compose trigger rejects compose files mounted outside app directory — Removed overly strict working-directory boundary enforcement from runComposeCommand that rejected compose files bind-mounted outside /home/node/app, breaking documented mount patterns like /drydock/docker-compose.yml. Compose file paths are operator-configured and already validated during resolution.
Compose trigger uses host paths instead of container paths — Docker label auto-detection (com.docker.compose.project.config_files) now remaps host-side paths to container-internal paths using Drydock's own bind mount information. Previously, host paths like /mnt/volume1/docker/stacks/monitoring/compose.yaml were used directly inside the container where they don't exist, causing "does not exist" errors even when the file was properly mounted.
Compose trigger logs spurious warnings for unrelated containers — When multiple compose triggers are configured, each trigger now silently skips containers whose resolved compose files don't match the trigger's configured FILE path, eliminating noisy cross-stack "does not exist" warnings.
Silent error on recheck failure — "Recheck for Updates" button now displays an error banner when the backend request fails instead of silently stopping the spinner with no feedback.
Silent error on env reveal failure — Environment variable reveal in the container detail panel now shows an inline error message when the API call fails instead of silently failing.
Security scans persist across navigation — Navigating away from the Security view no longer cancels in-flight batch scans. Module-scoped scan state survives unmount and the progress banner reappears on return.
SSE stale sweep timer on re-initialization — Stale client sweep interval now starts even when init() is called after a hot reload, preventing leaked SSE connections.
About modal documentation icon — Documentation link in the about modal now shows a book icon instead of the expand/maximize icon.
Log auto-fetch pauses in background tabs — useAutoFetchLogs now stops polling when the browser tab is hidden and automatically resumes when it becomes visible again.
SBOM download DOM isolation — Isolated DOM element creation and URL.createObjectURL references in the SBOM download composable, fixing potential memory leaks and test failures from uncleared object URLs. JSON serialization skipped when SBOM panel is hidden.
Dashboard resource fetch error propagation — Dashboard API fetch errors are now propagated consistently to the UI error state instead of being silently swallowed.
Docker image "latest" tag restricted to stable releases — CI release workflow no longer tags prerelease versions as latest, preventing unstable images from being pulled by default.
Security table refreshes progressively during batch scan — Security view table updates incrementally as each container scan completes instead of waiting for the entire batch to finish.
Vulnerability data fetched from per-container endpoint — Security view now fetches vulnerability data from the correct per-container endpoint instead of a missing bulk endpoint.
OIDC callback session loss with cross-site IdPs — Session cookies now default to SameSite=Lax for auth compatibility, fixing callback flows that could fail under SameSite=Strict. Added DD_SERVER_COOKIE_SAMESITE (strict|lax|none) for explicit control. (#52)
Compose trigger handles unknown update kinds — Containers with updateKind.kind === 'unknown' now trigger docker compose pull instead of silently skipping. (#91)
Compose image patching uses structured YAML edits — Replaced regex/indent heuristics with YAML parser targeting only services.<name>.image, preserving comments and formatting.
Hub/DHI public registries preserved with legacy token envs — Public registry fallback no longer lost when a private token is configured. Fail-closed behavior remains for private registry auth and runtime token exchange failures.
GHCR retries anonymously on credential rejection — Public image checks continue when configured credentials are rejected by GHCR/LSCR.
Partial registry registration failures isolated — Promise.allSettled prevents a single bad registry from taking down all registries including the public fallback.
Auth-blocked remote watchers stay registered — Remote watchers that fail auth now show as degraded instead of crashing watcher init.
Docker event stream reconnects with exponential backoff — Watcher reconnects automatically (1s doubling to 30s max) instead of staying disconnected after Docker socket interruption.
SSE frames flushed immediately — Added X-Accel-Buffering: no and explicit flush() to prevent nginx/traefik from buffering real-time events.
Store flushed on graceful shutdown — Explicit save() call on SIGTERM/SIGINT prevents data loss between autosave intervals.
Digest value populated on registration and refresh — Digest-watch containers no longer show undefined digest in the UI.
Icon fallback for missing upstream — Icon proxy returns bundled Docker fallback instead of 404 when upstream providers return 403/404. Fixes registry port parsing in icon URLs.
Container groups route no longer shadowed — /containers/groups mounted before /containers/:id to prevent Express treating group requests as container ID lookups.
Runtime env values redacted in API responses — Container environment variable values no longer exposed through the API.
Logger init failure produces structured stderr — Falls back to structured JSON on stderr instead of silent no-op when logger init fails.
Mobile sidebar closes on route change — Safety-net watcher ensures mobile menu closes on any navigation.
Security badge counts only scan vulnerabilities — No longer inflated by major version updates.
Trigger test failure shows parsed error message — Actionable error reason displayed below trigger card on test failure.
Viewport scrollbar eliminated — Fixed double-nested scroll contexts; long tags truncated with tooltips.
Self-hosted registries ignore port when matching — Registry matching now respects port numbers in self-hosted registry URLs, preventing mismatches between registries on different ports of the same host.
Socket proxy ECONNREFUSED with :ro mount — Removed :ro flag from Docker socket mount in all socket proxy compose examples. The read-only flag can prevent the proxy from connecting to the Docker daemon on some Linux kernels; the proxy's environment variable filtering (CONTAINERS=1, EVENTS=1, etc.) is the actual security boundary. Added health check and depends_on: condition: service_healthy to all socket proxy examples for proper startup ordering. Added FAQ entry for troubleshooting ECONNREFUSED with socket proxies.
Apprise trigger missing Content-Type header — Apprise notify requests now set explicit Content-Type: application/json header, fixing failures with Apprise servers that require it.
Toggle switch contrast — Improved toggle thumb contrast across all themes for better visibility.
Empty env var values rejected during container validation — Joi schema on details.env[].value used .string().required() which implicitly disallows empty strings. Containers with empty environment variable values (e.g. FOO=) were silently skipped during watch cycles. Fixed with .allow(''). (#120)
OIDC login broken by same-origin redirect check — LoginView.vue restricted OIDC redirects to same-origin URLs, but OIDC Identity Provider URLs are always cross-origin (Authentik, Keycloak, etc.). Removed the same-origin check, keeping only HTTP/HTTPS protocol validation.
Blank white screen on plain HTTP deployments — Helmet.js defaults enabled HSTS and upgrade-insecure-requests CSP even when TLS was not configured, causing browsers to block sub-resource loads on plain HTTP. Now conditionally omits upgrade-insecure-requests and HSTS when TLS is off. Strict boolean check on tls.enabled prevents string coercion. (#120)
Stale theme preferences after upgrade from v1.3 — Preferences migration deepMerge overwrote defaults with persisted values unconditionally, so invalid enum values (e.g. removed drydock theme family) survived migration and caused rendering failures. Added sanitize() pass that strips invalid theme families, variants, font families, icon libraries, scales, radius presets, view modes, and table actions before merge.
OIDC discovery fails on HTTP IdP URLs — openid-client v6 enforces HTTPS by default for all OIDC requests. Users running IdPs behind reverse proxies or on private networks with HTTP discovery URLs got "only requests to HTTPS are allowed" errors. Now passes allowInsecureRequests when the discovery URL protocol is http:.
Docker Compose trigger fails with EBUSY on bind-mounted stacks — writeComposeFileAtomic() used a single fs.rename() call that failed permanently when another process (e.g. Dockge) held the file or when Docker bind-mount overlay contention blocked the rename. Now retries up to 5 times with 200ms backoff, then falls back to a direct writeFile if rename remains blocked. (#84)
Drydock container display name hardcoded — The Docker watcher hardcoded drydock as the display name for drydock's own container instead of using the actual container name like every other container.
Non-existent DD_OIDC_ALLOW_HTTP env var referenced in UI — The UI OIDC HTTP banner referenced a DD_OIDC_ALLOW_HTTP env var that does not exist — the backend auto-detects http:// discovery URLs and passes allowInsecureRequests automatically. Removed the misleading reference.
Load test and start scripts broken by standardized API responses — jq queries in run-load-test.sh and start-drydock.sh used raw array syntax instead of .data[] to match the new collection response pattern.
Container status not reconciled on cron poll — Containers that changed state between Docker event stream updates (e.g. stopped externally) were not reconciled during periodic cron polls. Now reconciles container status on each poll cycle.
DD_AUTH_ANONYMOUS_CONFIRM env var alias rejected — DD_AUTH_ANONYMOUS_CONFIRM was not accepted as an alias for the canonical DD_ANONYMOUS_AUTH_CONFIRM env var, forcing users to use only the non-prefixed form.
LSCR cross-host auth failure — LinuxServer Container Registry (lscr.io) token exchange failed because the auth flow required cross-host authentication via the ghcr.io token endpoint. Now allows cross-host auth for lscr.io.
iPad sidebar clipping — Used dvh (dynamic viewport height) instead of vh for sidebar height to fix clipping on iPad and other mobile browsers where the toolbar reduces available viewport height.
Mobile dashboard scroll clipping — Dashboard content was clipped on mobile viewports when scrolling, preventing users from reaching all content.
Lockout counter reset on failed login — Brute-force lockout counter could reset prematurely, and strategy IDs were not protected from enumeration. Fixed counter persistence and added strategy ID protection.
Stale vulnerability data on fetch error — Security view retained stale vulnerability data when a fetch error occurred instead of clearing it, showing outdated information.
Audit service missing limit parameter — Audit service requests did not always send the limit query parameter, causing inconsistent pagination behavior.
Backup retention on failed updates — Backup entries are now pruned on the failure path, not just after successful updates, preventing indefinite accumulation of stale backups.
Backup pruning with undefined maxCount — pruneOldBackups() no longer deletes all backups when maxCount is undefined (e.g. when DD_BACKUPCOUNT is not configured). Now correctly no-ops on invalid or non-finite values.
Auto-rollback audit fromVersion accuracy — Rollback audit entries now correctly record fromVersion as the failing new image tag (via updateKind.remoteValue) instead of the pre-update old tag.
HASS entity_picture URL broken after logo rename — MQTT HASS discovery payload referenced a renamed logo file (drydock.png instead of whale-logo.png), causing missing entity pictures in Home Assistant. (#138)
Watcher crashes on containers with empty names — Docker watcher's same-name deduplication filter threw errors when containers had empty or missing names. Now skips deduplication for unnamed containers.
Container names not reconciled after external recreate — Containers recreated externally (via Portainer or docker compose up) retained stale names in the store until the next full poll cycle. Now reconciles container names immediately on detection.
Nested icon prefixes fail proxy request — Icon proxy rejected icons with doubled prefixes like mdi:mdi-docker. Now normalizes nested prefixes before proxying.
Colon-separated icon prefixes rejected — dd.display.icon labels using colon separators (e.g., sh:nextcloud) were rejected by the API validation pattern. Validation now accepts colon-prefixed icon identifiers.
Bouncer-blocked state missing from container details — Container detail views didn't reflect bouncer-blocked status. Now correctly wires the blocked state into detail panel display.

Security

Security API error sanitization — SBOM and scan error responses now return generic messages (Error generating SBOM, Security scan failed) instead of leaking internal error details. Detailed errors logged server-side only.
Agent log query parameter validation — Agent log level and component query parameters are validated against an allowlist and safe-character pattern, returning 400 for invalid values.
TLS key/cert paths stripped from config response — Server configuration API response no longer includes filesystem paths for TLS private key and certificate files.
Type-safe store and auth modules — Settings, notification, and auth store modules upgraded from any to explicit typed interfaces, preventing implicit type coercion vulnerabilities.
Fail-closed auth enforcement across registries and triggers — Bearer token, OIDC, and credential flows use failClosedAuth with typed RequestOptions, rejecting requests when required credentials are missing.
Input validation hardened — Additional input validation and error redaction across auth, API, and configuration modules.
Mutation-only JSON body parser — Express JSON body parsing restricted to mutation methods (POST/PUT/PATCH) only on both API and auth routers, reducing attack surface on read requests.
CSRF Sec-Fetch-Site validation — CSRF middleware now rejects requests with Sec-Fetch-Site: cross-site header, blocking cross-site state-changing requests even when the Origin header is absent.
HTTPS enforcement for SameSite=none cookies — DD_SERVER_COOKIE_SAMESITE=none now requires HTTPS configuration (DD_SERVER_TLS_ENABLED=true or DD_SERVER_TRUSTPROXY) and throws at startup if neither is set.
Remember-me endpoint requires authentication — /auth/remember POST moved after requireAuthentication middleware, preventing unauthenticated access.
Env reveal rate limit tightened — /api/containers/:id/env rate limit reduced from 100/min to 10/min to prevent credential enumeration. Server error responses return generic messages instead of internal details.
Trivy command path validation — Trivy binary paths are validated against shell metacharacters and path traversal before execution.
Digest scan cache LRU eviction — Scan result cache uses LRU eviction (max 500 entries, configurable via DD_SECURITY_SCAN_DIGEST_CACHE_MAX_ENTRIES) to prevent unbounded memory growth. Trivy DB status lookups are deduplicated across concurrent calls.
CSP configured for Iconify CDN — Content-Security-Policy updated to allow connect-src for the Iconify CDN origin, preventing blocked icon fetches in the browser.
CSP connect-src restricted in internetless mode — Content-Security-Policy connect-src directive tightened to 'self' when running in internetless mode, blocking outbound connections from the browser.
Legacy auth methods endpoint rate-limited — /api/auth/methods rate-limited to prevent enumeration of available authentication providers.
Removed plaintext credentials from login request body — The Basic auth login was redundantly sending username and password in both the Authorization header and the JSON body. The backend only reads the Authorization header via Passport, so the body credentials were unnecessary exposure.
Server-issued SSE client identity — Self-update ack requests validated against server-issued tokens, preventing spoofed acknowledgments.
Fail-closed auth across watchers, registries, and triggers — Token exchange failures no longer fall through to anonymous access.
Runtime env values redacted — Container environment variable values stripped from API responses to prevent credential leakage.
OIDC authorization redirect URL validation — Added allowlist-based validation for OIDC authorization redirect URLs, preventing open redirect attacks through crafted callback parameters.
Auth, registry token, and log sanitization hardening — Consolidated security pass hardening authentication flows, registry token validation, and log output sanitization.
Command trigger shell execution warning — Command trigger now logs a one-time security warning on first execution, reminding operators that commands run with drydock process privileges.
Login brute-force lockout — Per-account and per-IP lockout after configurable failed login attempts (DD_AUTH_ACCOUNT_LOCKOUT_MAX_ATTEMPTS, DD_AUTH_IP_LOCKOUT_MAX_ATTEMPTS) with configurable window and duration.
Concurrent session limits — Maximum authenticated sessions per user (default 5, configurable via DD_SERVER_SESSION_MAXCONCURRENTSESSIONS). Oldest sessions are revoked first when the limit is reached.
Destructive action confirmation header — Dangerous operations (delete container, restore backup, delete all containers) now require an X-DD-Confirm-Action header, returning 428 when missing.
Native argon2id password hashing — Replaced abandoned pass package with node:crypto argon2Sync for Basic auth. OWASP-aligned parameters with timing-safe comparison. Legacy {SHA} hashes accepted with deprecation warnings.
Full credential redaction — Component.mask() now returns [REDACTED] instead of leaking prefix/suffix characters in logs and API responses.
Trigger infrastructure config redaction — Webhook URLs, hostnames, channels, and usernames are redacted from trigger configuration in API responses.
Website SRI integrity hashes — Post-build script injects subresource integrity hashes for static assets on the documentation website.
Fail-closed webhook token enforcement — Per-endpoint webhook tokens now fail closed when a token is configured but the request provides no token or a mismatched token, preventing bypass through missing headers.
API error message sanitization — API routes no longer expose raw Joi validation or exception messages to clients. New sanitizeApiError() helper returns generic messages while logging real details server-side.
Trigger request body schema validation — POST /api/triggers/:type/:name and remote trigger endpoints now validate request bodies with Joi (require id string, reject type coercion via convert: false).
HTTP trigger auth schema enforcement at startup — HTTP trigger Joi schema now conditionally requires user+password for BASIC auth and bearer for BEARER auth at registration time, catching misconfigurations before first trigger execution.
CORS implicit wildcard origin deprecation warning — Startup warning when DD_SERVER_CORS_ENABLED=true without explicit DD_SERVER_CORS_ORIGIN. Default wildcard will require explicit opt-in in v1.5.0.
Identity-aware rate limit keying — Opt-in DD_SERVER_RATELIMIT_IDENTITYKEYING=true keys authenticated route rate limits by session/username instead of IP, preventing collisions for multiple users behind shared proxies. Unauthenticated routes remain IP-keyed. Disabled by default.
Reactive server feature flags in UI — Container action buttons (update, rollback, scan, triggers) are now gated by server-side feature flags via a useServerFeatures composable. When features like DD_SERVER_FEATURE_CONTAINERACTIONS are disabled, buttons show a disabled state with tooltip explaining why instead of silently failing at runtime.
Compose trigger hardening — Auto compose file detection from container labels (com.docker.compose.project.config_files) with Docker inspect fallback, pre-commit docker compose config --quiet validation before writes, compose file reconciliation (warn/block modes for runtime vs compose image drift), optional digest pinning (DIGESTPINNING trigger config), compose-file-once batch mode for multi-service stacks, multi-file compose chain awareness with deterministic writable target selection, compose metadata in update preview API, and compose file path display in container detail UI.
Unsupported hash formats fail closed — Basic auth now rejects unsupported hash formats instead of falling through to plaintext comparison, preventing accidental plaintext password acceptance.

Performance

Production source maps disabled — UI production builds now exclude source maps, reducing bundle size and improving deployment efficiency.
Audit store indexed date-range queries — Audit entries now store a pre-parsed timestampMs index for numeric comparisons. Date-range queries use the LokiJS chain API with indexed filtering instead of full-collection scans. Automatic 30-day retention with periodic pruning.
Backup store indexed lookups — Backup collection adds indices on data.containerName and data.id, using findOne() for single-document lookups and indexed find() for name-filtered queries instead of full scans.
LokiJS autosave interval set to 60 seconds — Fixed autosave interval at 60s instead of the LokiJS default, reducing disk I/O while maintaining acceptable data durability.
SSE shared heartbeat interval — Deduplicated per-client SSE heartbeat timers into a single shared interval that starts on first connection and stops when all clients disconnect.
LoginView exponential backoff — Login page connectivity retry uses exponential backoff (5s doubling to 30s max) instead of fixed intervals, reducing server load during outages.
Gzip response compression — API responses compressed above configurable threshold with automatic SSE exclusion.
Skip connectivity polling when SSE connection is active — Eliminates unnecessary /auth/user fetches every 10s during normal operation.
Set-based lookups replace linear scans — Repeated array lookups converted to Set operations in core paths.
Vite chunk splitting — Production build now splits vendor code into framework, icons, and vendor chunks for better browser cache efficiency across deployments.
Session index cache with atomic login locks — Auth session lookups use an indexed cache for O(1) access. Login operations use atomic locks to prevent race conditions during concurrent authentication attempts.
Proactive store cache eviction — LokiJS store proactively evicts stale cache entries before insertion, reducing memory pressure during high-throughput container watch cycles.
Async secret file loading — Secret file loading (DD_*__FILE env vars) uses fs/promises for non-blocking reads during startup configuration, improving initialization time with many secret files.

Dependencies

cron-parser v2 to v5 — Upgraded cron-parser from v2 to v5 (CronExpressionParser API). Note: joi-cron-expression still uses cron-parser v2 internally as its own dependency.
Removed pass package — Replaced abandoned pass password hashing with native node:crypto (covered under Security below).
Trivy upgraded to 0.69.3 — Bumped Trivy version in qlty configuration for latest vulnerability database and scanner improvements.

Deprecated

SHA-1 basic auth password hashes — Legacy {SHA}<base64> password hashes are accepted with deprecation warnings. SHA-1 support will be removed in v1.6.0. Migrate to argon2id hashing.

1.3.9 — 2026-02-22

Fixed

Release signing broken by cosign v3 API change — cosign sign-blob v3 silently ignores --output-signature and --output-certificate in keyless OIDC mode, producing an empty .sig file that fails upload. Release workflow now extracts signature and certificate from the cosign .bundle JSON as a fallback, handling both old (base64Signature/cert) and new (messageSignature.signature/verificationMaterial.certificate.rawBytes) bundle formats.
Shellcheck SC2086 in release signing step — Unquoted ${TAGS} expansion in container image signing replaced with read-loop into array to eliminate word-splitting/globbing risk.

Changed

CI and lefthook now run identical lint checks — CI lint job previously ran qlty check --filter biome (1 plugin) while lefthook ran qlty check (17 plugins). Both now run qlty check --all from the repo root, ensuring local pre-push catches exactly what CI catches.
Pre-commit hook auto-fixes lint issues — qlty check --fix runs on staged files at commit time, followed by a verify step. Lint drift no longer accumulates until push time.
Lefthook pre-push is sequential fail-fast — Switched from piped: false (parallel) to piped: true with priority ordering so failures surface immediately with clear output.

1.3.8 — 2026-02-22

Fixed

Docker Compose trigger silently no-ops for updateKind: unknown — When the update model classifies a change as unknown (e.g. created-date-only updates, unrecognized tag formats), getNewImageFullName resolved the update image identically to the current image, causing both compose-update and runtime-update filters to return empty arrays and log "All containers already up to date". The runtime-update filter now also triggers when container.updateAvailable === true, ensuring containers with confirmed updates are recreated regardless of updateKind classification. Compose file rewrites remain gated on explicit tag deltas. (#91)
Digest watch masks tag updates, pulling old image — When digest watch was enabled on a container with both a tag change and a digest change (e.g. v2.59.0-s6 → v2.60.0-s6), the update model gave digest unconditional priority, returning kind: 'digest' instead of kind: 'tag'. The trigger then resolved the image to the current tag (correct for digest-only updates) instead of the new tag, pulling the old image. Tag updates now take priority over digest when both are present. This bug was inherited from the upstream project (WUD). (#91)
Database not persisted on container shutdown — LokiJS relies on its autosave interval to flush data to disk, but the graceful shutdown handler called process.exit() before the next autosave tick could fire, causing any in-memory changes since the last autosave to be lost. This manifested as stale version numbers, lost update policies, and missing audit log entries after restarting the drydock container. Now explicitly saves the database during shutdown before exiting. This bug was inherited from the upstream project (WUD) but made deterministic by our graceful shutdown changes. (#96)

1.3.7 — 2026-02-21

Fixed

Tag regex OOM crash with re2-wasm — Replaced re2-wasm with re2js (pure JavaScript RE2 port). The WASM binary had a hard 16 MB memory ceiling with no growth allowed, causing abort() crashes on valid regex patterns like ^v(\d+\.\d+\.\d+)-ls\d+$. Since re2-wasm is abandoned (last npm publish Sep 2021) with no path to a fix, re2js provides the same linear-time ReDoS protection without WASM memory limits or native compilation requirements. (#89)
Self-signed/private CA support for self-hosted registries — Added optional CAFILE and INSECURE TLS options for self-hosted registry providers (Custom, Gitea, Forgejo, Harbor, Artifactory, Nexus). This allows private registries with internal or self-signed certificates to pass TLS validation via a mounted CA bundle, or to explicitly disable verification for trusted internal networks. (#88)
Docker Compose trigger silently no-ops on digest updates — Digest-only updates (same tag, new image hash) were filtered out entirely because the compose image string didn't change, causing the trigger to report success without recreating the container. Now digest updates skip the compose file write (correct — tag hasn't changed) but still trigger container recreation to pull the new image. (#91)

Changed

Gitea refactored to shared base class — Gitea now extends SelfHostedBasic directly instead of duplicating its logic from Custom, reducing code and ensuring consistent behavior with Harbor, Nexus, and Artifactory.
Lint tooling migrated from biome CLI to qlty — Removed @biomejs/biome as a direct devDependency from all workspaces; biome is now managed centrally via qlty. Lint and format scripts updated to use qlty check/qlty fmt.
Dependabot replaced with Renovate — Switched dependency update bot for better monorepo grouping, auto-merge of patch updates, and pinned GitHub Actions digests.
Socket Firewall switched to free mode — The CI supply chain scan now uses firewall-free (blocks known malware, no token required) instead of firewall-enterprise.
CI pipeline improvements — Added npm and Docker layer caching, parallelized e2e/load-test jobs, reordered job dependencies for faster feedback, added harden-runner to all workflow jobs.
CI credential hardening — Bumped harden-runner v2.11.1 → v2.14.2 (fixes GHSA-cpmj-h4f6-r6pq) and added persist-credentials: false to all actions/checkout steps across all workflows to prevent credential leakage through artifacts.
Zizmor added to local pre-push checks — GitHub Actions security linter now runs via qlty alongside biome, catching workflow misconfigurations before push.
Lefthook pre-push runs piped — Commands now run sequentially with fail-fast instead of parallel, so failures surface immediately instead of hanging while other commands complete.

1.3.6 — 2026-02-20

Fixed

GHCR anonymous auth returns 401 on public repos — The v1.3.3 fix for anonymous bearer tokens (Og==) removed the auth header entirely, but GHCR requires a token exchange even for unauthenticated pulls. Replaced direct bearer auth with proper token exchange via https://ghcr.io/token, matching the Hub/Quay pattern. Authenticated requests add Basic credentials to the token request; anonymous requests omit them. LSCR inherits the fix automatically. (#85, #86)

1.3.5 — 2026-02-19

Fixed

Container exits immediately when socket GID has no named group — Docker.entrypoint.sh treated getent group <gid> failures as fatal under set -e -o pipefail, so mounts where /var/run/docker.sock had a numeric GID not present in /etc/group caused an immediate exit (status=exited, exit=2) before app startup. The group lookup is now tolerant and falls back to creating a matching group as intended. (#82)
Log pretty-printing no longer depends on shell pipes — Moved human-readable formatting from the entrypoint pipeline (node | pino-pretty) into the app logger configuration. This preserves proper exec/signal behavior under tini while keeping DD_LOG_FORMAT=json support.

1.3.4 — 2026-02-19

Fixed

Backup lookup broken after container update — Backups were keyed by Docker container ID, which changes on every recreate (e.g. after an update). Switched all backup queries to use the stable container name, so backups are always found regardless of container ID changes. (#79)
Image prune deletes backup image — cleanupOldImages removed the previous image tag after updates, making rollback impossible. Now checks retained backup tags before pruning and skips images that are needed for rollback.
Auto-rollback monitor uses stale container ID — After an update recreates the container, maybeStartAutoRollbackMonitor passed the old (now-deleted) container ID to the health monitor. Now looks up the new container by name and passes the correct ID.
Backup stores internal registry name instead of Docker-pullable name — Backup imageName was stored as the internal registry-prefixed name (e.g. hub.public/library/nginx) which is not a valid Docker image reference. Rollback would fail with DNS lookup errors. Now stores the Docker-pullable base name (e.g. nginx) using the registry's getImageFullName method.
Rollback API docs incorrect endpoint — Fixed documentation showing /api/backup/:id/rollback instead of the correct /api/containers/:id/rollback.

1.3.3 — 2026-02-18

Fixed

Self-update leaves container stopped — When drydock updated its own container, stopping the old container killed the Node process before the new one could be created, leaving the UI stuck on "Restarting..." indefinitely. Now uses a helper container pattern: renames old container, creates new container, then spawns a short-lived helper that curls the Docker socket to stop old → start new → remove old. (#76)
Stale digest after container updates — After a container was updated (new image pulled, container recreated), the next watch cycle still showed the old digest because the early-return path in addImageDetailsToContainer skipped re-inspecting the Docker image. Now re-inspects the local image on each watch cycle to refresh digest, image ID, and created date. (#76)
express-rate-limit IPv6 key generation warning — Removed custom keyGenerator from the container scan rate-limiter that bypassed built-in IPv6 normalization, causing ERR_ERL_KEY_GEN_IPV6 validation errors.
express-rate-limit X-Forwarded-For warning — Added validate: { xForwardedForHeader: false } to all 6 rate-limiters to suppress noisy ERR_ERL_UNEXPECTED_X_FORWARDED_FOR warnings when running without trust proxy (e.g. direct Docker port mapping).
Quay auth token extraction broken — Fixed authenticate() reading response.token instead of response.data.token, causing authenticated pulls to silently run unauthenticated. Also affects Trueforge via inheritance.
GHCR anonymous bearer token — Fixed anonymous configurations sending Authorization: Bearer Og== (base64 of :) instead of no auth header, which could break public image access.
Created-date-only updates crash trigger execution — Fixed getNewImageFullName() crashing on .includes() of undefined when a container had only a created-date change (no tag change). Now rejects unknown update kind in threshold logic.
Compose write failure allows container updates — Fixed writeComposeFile() swallowing errors, allowing processComposeFile() to proceed with container updates even when the file write failed, causing runtime/file state desynchronization.
Self-update fallback removes running old container — Fixed helper script running removeOld after the fallback path (startOld), which would delete the running old container. Now only removes old after successful new container start.
Registry calls have no timeout — Added 30-second timeout to all registry API calls via Axios. Previously a hung registry could stall the entire watch cycle indefinitely.
HTTP trigger providers have no timeout — Added 30-second timeout to all outbound HTTP trigger calls (Http, Apprise, Discord, Teams, Telegram). Previously a slow upstream could block trigger execution indefinitely.
Kafka producer connection leak — Fixed producer connections never being disconnected after send, leaking TCP connections to the broker over time. Now wraps send in try/finally with disconnect.
Rollback timer labels not validated — Invalid dd.rollback.window or dd.rollback.interval label values (NaN, negative, zero) could cause setInterval to fire continuously. Now validates with Number.isFinite() and falls back to defaults.
Health monitor overlapping async checks — Added in-flight guard to prevent overlapping health checks from triggering duplicate rollback executions when inspections take longer than the poll interval.
Anonymous login double navigation guard — Fixed beforeRouteEnter calling next() twice when anonymous auth was enabled, causing Vue Router errors and nondeterministic redirects.
Container API response not validated — Fixed getAllContainers() not checking response.ok before parsing, allowing error payloads to be treated as container arrays and crash computed properties.

Security

fast-xml-parser DoS via entity expansion — Override fast-xml-parser 5.3.4→5.3.6 to fix CVE GHSA-jmr7-xgp7-cmfj (transitive dep via @aws-sdk/client-ecr, upstream hasn't released a fix yet).
tar arbitrary file read/write — Removed tar from dependency graph entirely by replacing native re2 (which pulled in node-gyp → tar) with re2-wasm (v1.3.3), later replaced by re2js (v1.3.7) due to WASM memory limits. Previously affected by CVE GHSA-83g3-92jg-28cx.
Unauthenticated SSE endpoint — Moved /api/events/ui behind requireAuthentication middleware and added per-IP connection limits (max 10) to prevent connection exhaustion.
Session cookie missing sameSite — Set sameSite: 'strict' on session cookie to mitigate CSRF attacks.
Predictable session secret — Added DD_SESSION_SECRET environment variable override so deployments can provide proper entropy instead of the default deterministic UUIDv5.
Global error handler leaks internal details — Replaced err.message with generic 'Internal server error' in the global error handler to prevent leaking hostnames, paths, and Docker socket info to unauthenticated callers.
Entrypoint masks crash exit codes — Enabled pipefail in Docker.entrypoint.sh so node | pino-pretty correctly propagates non-zero exit codes for restart policies.

1.3.2 — 2026-02-16

Added

Log viewer auto-fetch polling — Configurable auto-fetch interval (Off / 2s / 5s / 10s / 30s) for both application and container log viewers, replacing manual-only refresh. Defaults to 5 seconds for a near-real-time tail experience. (#57)
Log viewer scroll lock — Scrolling away from the bottom pauses auto-scroll, showing a "Scroll locked" indicator and "Resume" button. New log data continues to load in the background without yanking the user's scroll position. (#57)
Log viewer auto-scroll — New log entries automatically scroll the view to the bottom when the user is near the end, providing a tail-like experience. (#57)
Shared log viewer composable — Extracted useLogViewerBehavior composable with useLogViewport (scroll management) and useAutoFetchLogs (interval timer lifecycle) to eliminate duplication between application and container log views.
7 new registry providers — Added OCIR (Oracle Cloud), IBMCR (IBM Cloud), ALICR (Alibaba Cloud), GAR (Google Artifact Registry), Harbor, JFrog Artifactory, and Sonatype Nexus. Includes a shared SelfHostedBasic base class for self-hosted registries with basic auth.
4 new trigger providers — Added Mattermost, Microsoft Teams (Adaptive Cards), Matrix, and Google Chat notification triggers.

Fixed

v1 manifest digest watch using image ID instead of repo digest — Fixed handleDigestWatch() incorrectly reading Config.Image (the local image ID) as the digest for v1 manifest images, causing perpetual false "update available" notifications. Now uses the repo digest from RepoDigests instead. (getwud/wud#934)
Discord trigger broken after request→axios migration — Fixed sendMessage() using request-style properties (uri, body) instead of axios properties (url, data), causing "Invalid URL" errors on all Discord webhook calls. (getwud/wud#933)

1.3.1 — 2026-02-15

Fixed

Release SBOM generation for multi-arch images — Replaced anchore/sbom-action (which fails on manifest list digests from multi-platform builds) with Docker buildx native SBOM generation (sbom: true), producing per-platform SBOMs embedded in image attestations.

Security

Pin Trivy install script by commit hash — Replaced mutable main branch reference in Dockerfile curl | sh with a pinned commit SHA to satisfy OpenSSF Scorecard pinned-dependencies check and prevent supply-chain risk from upstream changes.

1.3.0 — 2026-02-15

Fixed

OIDC session resilience for WUD migrations — Corrupt or incompatible session data (e.g. from WUD's connect-loki store) no longer causes 500 errors. Sessions that fail to reload are automatically regenerated. All OIDC error responses now return JSON instead of plain text, preventing frontend parse errors. Added a global Express error handler to ensure unhandled exceptions return JSON.
Disabled X-Powered-By header — Removed the default Express X-Powered-By header from both the main API and agent API servers to reduce information exposure.
Trivy scan queue — Serialized concurrent Trivy invocations to prevent "cache may be in use by another process" errors when multiple containers are scanned simultaneously (batch triggers, on-demand scans, SBOM generation).
Login error on wrong password — loginBasic() attempted to parse the response body as JSON even on 401 failures, causing Unexpected token 'U', "Unauthorized" is not valid JSON errors instead of the friendly "Username or password error" message.
Snackbar notification colors ignoring level — The SnackBar component had a hardcoded color="primary" instead of binding to the level prop, causing error and warning notifications to display as blue instead of red/amber.
SBOM format key mismatch — Fixed container model schema validating SBOM formats against cyclonedx instead of the correct cyclonedx-json key.

Added

Snyk vulnerability monitoring — Integrated Snyk for continuous dependency scanning of app/package.json and ui/package.json. Added Snyk badge to README with targetFile parameter for monorepo support.
Update Bouncer (Trivy safe-pull gate) — Added pre-update vulnerability scanning for Docker-triggered updates. Candidate images are scanned before pull/restart, updates are blocked when vulnerabilities match configured blocking severities, and latest scan data is persisted on container.security.scan. Added GET /api/containers/:id/vulnerabilities endpoint for retrieving scan results.
Update Bouncer signature verification (cosign) — Added optional pre-update image signature verification. When enabled, Docker-triggered updates are blocked if candidate image signatures are missing/invalid or verification fails.
Update Bouncer SBOM generation — Added Trivy SBOM generation (spdx-json, cyclonedx-json) for candidate images with persistence in container.security.sbom and a new GET /api/containers/:id/sbom API endpoint (with format query support).
Container card security status chip — Added a vulnerability chip on container cards showing Update Bouncer scan status (safe, blocked, scan error) with severity summary tooltip data from container.security.scan.
On-demand security scan — Added POST /api/containers/:id/scan endpoint for triggering vulnerability scan, signature verification, and SBOM generation on demand. Broadcasts dd:scan-started and dd:scan-completed SSE events for real-time UI feedback. Added shield button to container card actions and mobile overflow menu.
Direct container update from UI — Added POST /api/containers/:id/update endpoint that triggers a Docker update directly without requiring trigger configuration. The "Update now" button in the UI now calls this single endpoint instead of looping through configured triggers.
Trivy and cosign in official image — The official drydock image now includes both trivy and cosign binaries, removing the need for custom images in local CLI mode.

Changed

README badge layout — Added line breaks to badge rows for a cleaner two-line layout across all three badge sections.
Grafana dashboard overhaul — Updated overview dashboard with standard datasource naming (DS_PROMETHEUS), added bar chart and pie chart panels, and restructured panel layout for better monitoring coverage.
Mobile responsive dashboard — Stat cards now stack full-width on small screens with tighter vertical spacing for a cleaner mobile layout.
Self-update overlay rendering — Switched logo images from v-if to v-show to avoid re-mount flicker during self-update phase transitions.
Container sort simplification — Simplified null-group sorting in ContainersView using sentinel value instead of multi-branch conditionals.
Test coverage improvements — Expanded app test coverage for API routes (backup, container-actions, preview, webhook), OIDC authentication, registry component resolution, tag parsing, and log sanitization. Expanded UI test coverage across 38 spec files with improved Vuetify stub fidelity (v-tooltip activator slot, v-list-item slots, app-bar-nav-icon events).
Vitest coverage config — Narrowed coverage to .js/.ts files only (excluding .vue SFCs) to avoid non-actionable template branch noise.
Prometheus counter deduplication — Extracted shared createCounter factory in app/prometheus/counter-factory.ts, reducing boilerplate across audit, webhook, trigger, and container-actions counter modules.
API error handler deduplication — Extracted shared handleContainerActionError helper in app/api/helpers.ts, consolidating duplicate catch-block logic across backup, preview, and container-actions routes.
Lint and code quality fixes — Fixed biome noPrototypeBuiltins warning in OIDC tests, added id attributes to README HTML headings to resolve markdownlint MD051, and tuned qlty smell thresholds.

Security

CodeQL alert fixes — Fixed log injection vulnerabilities by sanitizing user-controlled input before logging. Removed unused variables flagged by static analysis. Added rate limiting to the on-demand scan endpoint.
Build provenance and SBOM attestations — Added supply chain attestations to release workflow for verifiable build provenance.

1.2.0

Added

Grafana dashboard template — Importable Grafana JSON dashboard with panels for overview stats, watcher activity, trigger execution, registry response times, and audit entries. Uses datasource templating for portable Prometheus configuration.
Audit log backend — AuditEntry model, LokiJS-backed store with pagination and pruning, GET /api/audit endpoint with filtering, dd_audit_entries_total Prometheus counter, and automatic logging of container lifecycle events (update-available, update-applied, update-failed, rollback, preview, container-added, container-removed).
Font Awesome 6 migration — Replaced all Material Design Icons (mdi-*) with Font Awesome 6 equivalents. Configured Vuetify FA icon set, updated all service icon getters, component templates, and 54 test files.
Dry-run preview API — POST /api/containers/:id/preview returns what an update would do (current/new image, update kind, running state, networks) without performing it.
Pre-update image backup and rollback — LokiJS-backed backup store records container image state before each Docker trigger update. GET /api/backups, GET /api/:id/backups, and POST /api/:id/rollback endpoints. Configurable retention via DD_TRIGGER_DOCKER_{name}_BACKUP_COUNT (default 3).
Frontend wiring — Preview dialog with loading/error/success states wired to dry-run API. Full audit log table with filtering, pagination, and responsive column hiding replacing the MonitoringHistory placeholder. Recent Activity dashboard card showing latest 5 audit entries.
Container action bar refactor — Replaced 3-column text button layout with compact icon-button toolbar and tooltips (desktop) or overflow menu (mobile).
Dashboard second row — Added Recent Activity and stats cards as a second row on the dashboard.
UI modernization — Consistent pa-4 padding, outlined/rounded cards, tonal chips, styled empty states, and Font Awesome icons across all views and components.
Container actions (start/stop/restart) — New API endpoints and UI buttons to start, stop, and restart Docker containers directly from the dashboard. Gated by DD_SERVER_FEATURE_CONTAINERACTIONS (default: enabled). Includes audit logging, Prometheus counter (dd_container_actions_total), desktop toolbar buttons with disabled-state awareness, and mobile overflow menu integration.
Webhook API for on-demand triggers — Token-authenticated HTTP endpoints (POST /api/webhook/watch, /watch/:name, /update/:name) for CI/CD integration. Gated by DD_SERVER_WEBHOOK_ENABLED and DD_SERVER_WEBHOOK_TOKEN. Includes rate limiting (30 req/15min), audit logging, Prometheus counter (dd_webhook_total), and a configuration info panel on the Server settings page.
Container grouping / stack views — New GET /api/containers/groups endpoint returns containers grouped by stack. Supports explicit group assignment via dd.group / wud.group labels with automatic fallback to com.docker.compose.project. Collapsible ContainerGroup component with group header showing name, container count, and update badges. "Smart group" filter option for automatic stack detection (dd.group > wud.group > compose project). "Update all in group" action to batch-update all containers in a group.
Graceful self-update UI — Self-update detection when drydock updates its own container. Server-Sent Events (SSE) endpoint at /api/events/ui for real-time browser push. Full-screen DVD-style bouncing whale logo overlay during self-updates with smooth phase transitions (updating, restarting, reconnecting, ready). Automatic health polling and page reload after restart.
Lifecycle hooks (pre/post-update commands) — Execute shell commands before and after container updates via dd.hook.pre and dd.hook.post labels. Pre-hook failures abort the update by default (dd.hook.pre.abort=true). Configurable timeout via dd.hook.timeout (default 60s). Environment variables exposed: DD_CONTAINER_NAME, DD_IMAGE_NAME, DD_TAG_OLD, DD_TAG_NEW, etc. Includes audit logging for hook success/failure and UI display in ContainerDetail panel.
Automatic rollback on health check failure — Monitors container health after updates and automatically rolls back to the previous image if the container becomes unhealthy. Configured via dd.rollback.auto=true, dd.rollback.window (default 300s), and dd.rollback.interval (default 10s). Requires Docker HEALTHCHECK on the container. Uses existing backup store for rollback images. Includes audit logging and UI display in ContainerDetail panel.
selfhst/icons as primary icon CDN — Switched to selfhst/icons as the primary icon CDN with homarr-labs as fallback, improving icon availability and coverage.

Fixed

Navigation drawer not visible — Used computed model for permanent/temporary modes; passing model-value=undefined caused Vuetify to treat the drawer as closed.
Dark theme missing colors — Added info, success, and warning color definitions to the dark theme.
ContainerPreview updateKind display — Fixed structured updateKind object rendering with semver-diff color coding.
Invalid text-body-3 CSS class — Replaced with valid text-body-2 in ConfigurationItem and TriggerDetail.
404 catch-all route — Added catch-all redirect to home for unknown routes.
False downgrade suggestion for multi-segment tags — Fixed semver parsing/comparison for numeric tags like 25.04.2.1.1 so newer major tags are no longer suggested as downgrades. (#47)
Configured path hardening for filesystem reads — Added validated path resolution helpers and applied them to store paths, watcher TLS files, and MQTT TLS files before filesystem access.

Changed

Audit event wiring — Wired audit log entries and Prometheus counter increments for rollback, preview, container-added, container-removed, update-applied, and update-failed events. Registered ContainerUpdateFailed event with try/catch in Docker trigger.
Test updates — 20+ test files updated for v1.2.0 icon changes, CSS selectors, HomeView data model, theme toggle relocation, and audit module wiring. Removed obsolete specs.
Updated doc icon examples — Switched icon examples to prefer hl: and si: prefixes over deprecated mdi:.
Code quality tooling consolidation — Replaced Codacy + SonarCloud with Qlty + Snyk. Rewrote lefthook.yml pre-push hooks to run qlty check, snyk test, snyk code test (informational), builds, and tests. Added scripts/snyk-code-gate.sh wrapper.
Biome formatting — Applied biome format across entire codebase for consistent code style.
README badges — Replaced Codacy/SonarCloud badges with CI status, Qlty maintainability, and Snyk badges.
ConfigurationItem redesign — Icon moved to the left with name as prominent text and type as subtitle, replacing the old badge/chip pattern across all configuration pages.
TriggerDetail redesign — Same modern layout treatment as ConfigurationItem (icon left, name prominent, type subtitle).
Registry page brand colors — Added brand-colored icon backgrounds for each registry provider (Docker blue, GitHub purple, AWS orange, Google blue, etc.) via getRegistryProviderColor() helper and new iconColor prop on ConfigurationItem.
Consistent card styling — Unified variant="outlined" rounded="lg" across ContainerItem, ContainerGroup, ContainerTrigger, and WebhookInfo cards for a cohesive look.
Home page severity badges removed — Removed redundant MAJOR/MINOR severity badges from the container updates list; version chip color already indicates severity.
History page filter bar — Removed redundant "Update History" heading (already in app bar) and added a collapsible filter bar with active filter chips.
Logs page spacing — Fixed spacing between the config item and logs card.
Self-update overlay responsive — Mobile-responsive self-update overlay uses static top-center positioning with fade-in animation on small screens instead of DVD bounce.
QA compose enhancements — Added HTTP trigger, basic auth, and webhook configuration to test/qa-compose.yml for integration testing.
Login page redesign — Redesigned login page with new font, icon colors, and layout polish.
Docker Hub and Quay.io multi-registry publishing — Container images now published to Docker Hub and Quay.io alongside GHCR for broader registry availability.
Mobile responsive dashboard — Per-type colored update badges (major=red, minor=warning, patch=success, digest=info) and icon-only tabs on mobile viewports.
Dark mode app bar logo inversion — App bar logo now inverts correctly in dark mode for improved visibility.
History page mobile improvements — Shorter timestamps, hidden status column, and truncated container names on mobile viewports.
Container filter mobile labels — Short labels ("Updates", "Time") on mobile breakpoint for compact filter display.
Biome and Qlty config alignment — Aligned Biome and Qlty configurations for consistent code quality enforcement.

Security

RE2 regex engine — Replaced native RegExp with Google's RE2 (re2 npm package) for all user-supplied regex patterns (includeTags, excludeTags, transformTags). RE2 uses a linear-time matching algorithm that is inherently immune to ReDoS catastrophic backtracking.
Docs dependency vulnerability fixes — Fixed 9 CVEs in docs/ transitive dependencies via npm overrides (dompurify 2→3, marked 1→4, got 9→11).

Removed

Dead code removal — Deleted unused AppFooter and ConfigurationStateView components, dead computed props (filteredUpdates, upToDateCount), duplicate isTriggering reset, dead mdi: prefix replacement in IconRenderer, dead container-deleted listener, and Maintenance Windows placeholder.
Removed @mdi/font dependency — Dropped unused Material Design Icons package.
Removed Codacy and SonarCloud — Replaced with Qlty (local code quality) and Snyk (dependency + SAST scanning) for a unified local-first quality gate.
Removed stale tracking docs — Deleted SONARQUBE-ISSUES.md, docs/sonar-smells-tracking.md, and docs/codacy-high-findings-tracking.md.

Documentation

Popular imgset presets — Added a curated preset guide at docs/configuration/watchers/popular-imgsets.md and linked it from watcher docs.

1.1.3

Fixed

ERR_ERL_PERMISSIVE_TRUST_PROXY on startup — Express trust proxy was hard-coded to true, which triggers a validation error in express-rate-limit v8+ when the default key generator infers client IP from X-Forwarded-For. Replaced with a configurable DD_SERVER_TRUSTPROXY env var (default: false). Set to 1 (hop count) when behind a single reverse proxy, or a specific IP/CIDR for tighter control. (#43)

1.1.2

Fixed

Misleading docker-compose file error messages — When a compose file had a permission error (EACCES), the log incorrectly reported "does not exist" instead of "permission denied". Now distinguishes between missing files and permission issues with actionable guidance. (#42)
Agent watcher registration fails on startup — Agent component path resolved outside the runtime root (../agent/components instead of agent/components), causing "Unknown watcher provider: 'docker'" errors and preventing agent watchers/triggers from registering. (#42)

Changed

Debug logging for component registration — Added debug-level logging showing resolved module paths during component registration and agent component registration attempts, making path resolution issues easier to diagnose.

1.1.1 - 2026-02-11

Fixed

Read-only Docker socket support — Drydock's privilege drop prevented non-root users from connecting to :ro socket mounts. Added DD_RUN_AS_ROOT=true env var to skip the drop, improved EACCES error messages with actionable guidance, and documented socket proxy as the recommended secure alternative. (#38)
Prometheus container gauge crash with agent containers — The container gauge used a blacklist filter that let unknown properties (like agent) slip through and crash prom-client. Switched to a whitelist of known label names so unknown properties are silently ignored. (#39)
Snackbar toast transparency — Used flat variant for solid background on toast notifications.
Container filter layout broken on narrow viewports — Filter columns rendered text vertically when the nav drawer was open because all 8 v-col elements had no width constraints. Added responsive breakpoints (cols/sm/md) so filters wrap properly across screen sizes. (#40)

1.1.0 - 2026-02-10

Added

Application log viewer — New Configuration > Logs page with a terminal-style viewer for drydock's own runtime logs (startup, polling, registry checks, trigger events, errors). Backed by an in-memory ring buffer (last 1,000 entries) exposed via GET /api/log/entries. Supports level filtering (debug/info/warn/error), configurable tail count (50/100/500/1,000), color-coded output, and auto-scroll to newest entries. An info tooltip shows the configured server log level.
Agent log source selector — When agents are configured, a "Source" dropdown appears in the log viewer to switch between the controller's own logs and any connected agent's logs. Disconnected agents are shown but disabled. Agent logs are proxied via GET /api/agents/:name/log/entries.
Container log viewer — New "Logs" tab in the container detail expansion panel to view container stdout/stderr output directly in the UI with tail control and refresh.

1.0.2 - 2026-02-10

Fixed

Registry and trigger crashes in agent mode — getSummaryTags() and getTriggerCounter() also return undefined in agent mode. Added optional chaining to all remaining Prometheus call sites so agent mode doesn't crash when processing containers or firing triggers. (Fixes #33)

1.0.1 - 2026-02-10

Fixed

Prometheus gauge crash in agent mode — getWatchContainerGauge() returns undefined in agent mode since Prometheus is not initialized. Added optional chaining so the .set() call is safely skipped. This was the root cause of containers not being discovered in agent mode. (Fixes #23, #31)

Changed

su-exec privilege dropping — Entrypoint detects the docker socket GID and drops from root to the node user via su-exec when possible. Stays root only for GID 0 sockets (Docker Desktop / OrbStack). (Refs #25)
tini init system — Added tini as PID 1 for proper signal forwarding to the Node process.
Graceful shutdown — SIGINT/SIGTERM handlers now call process.exit() after cleanup so the container actually stops.

1.0.0 - 2026-02-10

First semver release. Drydock adopts semantic versioning starting with this release, replacing the previous CalVer (YYYY.MM.PATCH) scheme.

Security

ReDoS prevention — Replaced vulnerable regexes in trigger template evaluation (Trigger.ts) with linear-time string parsing (parseMethodCall, isValidPropertyPath). Added MAX_PATTERN_LENGTH guards in tag transform (tag/index.ts) and Docker watcher (Docker.ts) to reject oversized user-supplied regex patterns.
XSS prevention — Added escapeHtml() sanitizer to Telegram trigger bold() method, preventing HTML injection via container names or tag values.
Workflow hardening — Set top-level permissions: read-all in release.yml and codeql.yml. Pinned all CodeQL action refs to commit hashes. Added CodeQL config to exclude js/clear-text-logging false positives.
CVE-2026-24001 — Updated diff dependency in e2e tests (4.0.2 → 4.0.4).

Changed

+285 UI tests — 15 new spec files and 7 expanded existing specs covering configuration views, container components, trigger detail, services, router, and app shell. UI test count: 163 → 285.
+59 app tests — New edge-case tests for ReDoS guard branches, parseMethodCall parsing, and Docker watcher label resolution. App test count: 1,254 → 1,313.
Complexity refactors — Extracted helpers from high-complexity functions: parseTriggerList/applyPolicyAction (container.ts), resolveLabelsFromContainer/mergeConfigWithImgset (Docker.ts).
Biome lint fixes — import type corrections and unused variable cleanup across 17 files.
Fixed doc links — Corrected broken fragment links in docs/_coverpage.md.

Removed

Removed legacy vue.config.js — Dead Vue CLI config file; project uses Vite.

2026.2.3 - 2026-02-10

Fixed

NTFY trigger auth 401 — Bearer token auth used unsupported axios.auth.bearer property; now sends Authorization: Bearer <token> header. Basic auth property names corrected to username/password. (#27)
Agent mode missing /health — Added unauthenticated /health endpoint to the agent server, mounted before the auth middleware so Docker healthchecks work without the agent secret. (#27)

Changed

Lefthook pre-push hooks — Added lefthook.yml with pre-push checks (lint + build + test).
Removed startup warning — Removed "Known Issue" notice from README now that container startup issues are resolved.

2026.2.2 - 2026-02-10

Security

Cosign keyless signing — Container image releases are now signed with Sigstore cosign keyless signing for supply chain integrity.
Least-privilege workflow permissions — Replaced overly broad read-all with minimum specific permissions across all CI/CD workflows.
CodeQL and Scorecard fixes — Resolved all high-severity CodeQL and OpenSSF Scorecard security alerts.
Pinned CI actions — All CI action references pinned to commit hashes with Dockerfile base image digest.

Added

Auto-dismiss notifications after container update — New resolvenotifications option for triggers (default: false). When enabled, notification triggers automatically delete the sent message after the Docker trigger successfully updates the container. Implemented for Gotify via its deleteMessage API. Other providers (Slack, Discord, ntfy) can add support by overriding the new dismiss() method on the base Trigger class. New containerUpdateApplied event emitted by the Docker trigger on successful update.

Fixed

Agent mode Prometheus crash — Guard getWatchContainerGauge().set() against undefined in Agent mode where Prometheus is not initialized, fixing "Cannot read properties of undefined (reading 'set')" crash (#23)
Sanitize version logging — Sanitize version strings from env vars before logging to resolve CodeQL clear-text-logging alerts in index.ts and store/migrate.ts
Broken event test assertion — Fix expect() without matcher in event test

Changed

97% test coverage — Boosted from 76% to 97% with 449 new tests (1,254 total across 95 test files).
Fuzz testing — Added property-based fuzz tests with fast-check for Docker image name parsing.
Static analysis fixes — Optional chaining, String#replaceAll(), readonly modifiers, Number.NaN, concise regex syntax, removed unused imports, moved functions to outer scope.
Reduced code duplication — Refactored duplicated code in registries, triggers, and store test files flagged by SonarCloud.
Pino logging — Replaced bunyan with pino to eliminate vulnerable transitive dependencies. Added pino-pretty for human-readable log output.
Renamed wud to drydock — Project references updated from upstream naming across Dockerfile, entrypoint, package files, scripts, and test fixtures.
CONTRIBUTING.md — Added contributor guidelines.
OpenSSF Best Practices badge — Added to README.
SonarCloud integration — Added project configuration.
Multi-arch container images — Docker images now built for both linux/amd64 and linux/arm64 architectures, published to GHCR.
Lefthook pre-push hooks — Added lefthook config with pre-push checks (lint + build + test) and npm run check convenience script.
CodeQL query exclusion — Exclude js/clear-text-logging query (false positives on DD_VERSION env var).

2026.1.0

Added

Agent mode — Distributed monitoring with remote agent architecture. Agent components, SSE-based communication, dedicated API routes.
OIDC token lifecycle — Remote watcher HTTPS auth with Basic + Bearer token support. TLS/mTLS compatibility for DD_WATCHER_{name}_HOST.
OIDC device-flow (Phase 2) — RFC 8628 Device Authorization Grant for headless remote watcher auth. Auto-detection, polling with backoff, and refresh token rotation.
Per-image config presets — imgset defaults for per-image configuration. Added watchDigest and inspectTagPath imgset properties.
Hybrid triggers — Trigger group defaults (DD_TRIGGER_{name}_THRESHOLD) shared across providers. Name-only include/exclude for multi-provider trigger management.
Container update policy — Skip/snooze specific update versions. Per-container policy stored in DB, exposed via API and UI.
Metrics auth toggle — DD_SERVER_METRICS_AUTH env var to disable auth on /metrics endpoint.
Trigger thresholds — Digest and no-digest thresholds for triggers.
NTFY provider-level threshold — Provider-level threshold support for ntfy trigger.
Docker pull progress logging — Rate-limited pull progress output during docker-compose updates.
Registry lookup image override — lookupImage field on registry config to override the image used for tag lookups.
Docker inspect tag path — Support custom tag path in Docker inspect output.
Anonymous LSCR and TrueForge registries — Allow anonymous access to LSCR (LinuxServer) and Quay-backed TrueForge.
DHI registry — New dhi.io registry provider with matcher, auth flow, and docs.
Custom URL icons — Support URL-based icons via dd.display.icon label.
Version skip — Skip specific versions in the UI.
Log viewer — In-app container log viewer. View Docker container stdout/stderr output directly in the UI via a new "Logs" tab on each container. Supports configurable tail line count (50/100/500), manual refresh, and Docker stream demultiplexing. Works for both local and remote agent containers.
Semver tag recovery — Recover include-filter mismatched semver tags from watchers. Extended to advise best semver tag when current tag is non-semver (e.g., latest).
Dashboard update chips — Replaced verbose update status text with compact colored chips: green "up to date" or warning "N update(s)" (clickable).

Fixed

eval() code injection — Replaced eval() in trigger template rendering with safe expression evaluator supporting property paths, method allowlist, ternaries, and string concatenation.
Digest-only update prune crash — Docker trigger prune logic now correctly excludes current image during digest-only updates and handles post-prune errors gracefully.
Swarm deploy-label debug logging — Added warn-level logging when Swarm service inspect fails, and debug logging showing which label sources contain dd.* labels.
OIDC session state races — Serialized redirect session checks, multiple pending callback states per session.
semverDiff undefined — Normalized semverDiff for non-tag (digest-only/created-date-only) updates.
Docker event stream crash — Buffered and parsed split Docker event stream payloads.
Multi-network container recreate — Reconnects additional networks after container recreation.
Remote watcher delayed first scan — watchatstart now checks watcher-local store for new remote watchers.
docker-compose post_start hooks — Hooks now execute after updates.
docker-compose image-only triggers — Only trigger on compose services with actual image changes.
docker-compose imageless services — Skip compose services without an image field.
docker-compose implicit latest tag — Normalize image: nginx to image: nginx:latest so compose triggers don't treat implicit latest as a version mismatch.
Express 5 wildcard routes — Named wildcard route params for express 5 compatibility.
Semver filtering — Fixed semver part filtering and prefix handling.
SMTP TLS_VERIFY inverted — rejectUnauthorized was inverted; TLS_VERIFY=false now correctly allows self-signed certificates.
HA MQTT deprecated object_id — Replaced object_id with default_entity_id for Home Assistant 2025.10+ compatibility.
Open redirect on authenticated pages — Validate next query parameter to only allow internal routes.
Trigger test updateKind crash — Test-button triggers no longer crash with "Cannot read properties of undefined (reading 'updateKind')" on unvalidated containers.
Docker rename event not captured — Added rename to Docker event listener so container name updates are captured after compose recreates.
UI duplicate drawer logo — Removed duplicate logo in navigation drawer.

Changed

TypeScript migration (app) — Entire backend converted from JavaScript to TypeScript with ES Modules (NodeNext). 232 .ts files added/renamed, all .js source files removed.
TypeScript migration (UI) — Vue 3 frontend migrated from JS to TS. 29 .vue files updated, component props/emits typed.
Jest → Vitest (app) — All 64 app test files (664 tests) migrated from Jest to Vitest. Test runner unified across app and UI.
Jest → Vitest (UI) — UI unit tests migrated from Jest to Vitest with improved coverage.
Vitest 4 + modern deps — Upgraded vitest 3→4, uuid 11→13, flat 5→6, snake-case 3→4. Fixed vitest 4 mock constructor breaking change.
ESM baseline — Cut over to NodeNext module resolution. Removed Babel, added tsconfig.json.
Biome linter — Replaced ESLint with Biome for formatting and linting.
CI cleanup — Removed Code Climate config, renamed Travis config to ci.config.yml.

Dependencies

Package	Upstream (8.1.1)	drydock
vitest	3.x (Jest)	4.x
uuid	9.x	13.x
flat	5.x	6.x
snake-case	3.x	4.x
express	4.x	5.x
typescript	—	5.9
biome	—	2.3

Stats: 392 files changed, +25,725 insertions, -25,995 deletions, 872 total tests (709 app + 163 UI).

Upstream Backports

The following changes from upstream/main (post-fork) have been ported to drydock:

Description	Status
Add Codeberg to default registries	Ported (new TS provider)
Increase `maxAliasCount` in YAML parsing	Ported
Fix authentication for private ECR registry (async `getAuthPull`)	Ported across all registries
Prometheus: add `DD_PROMETHEUS_ENABLED` config	Ported
Fix Authelia OIDC docs (field names)	Ported
Buffer Docker event stream before JSON parse	Already fixed independently
SMTP trigger: allow display name in from address (#908)	Ported

Remaining upstream-only changes (not ported — not applicable to drydock):

Description	Reason
Fix e2e tests (x2)	JS-based, drydock tests are TS
Fix prettier	drydock uses Biome
Fix codeberg tests	Covered by drydock's own tests
Update changelog	Upstream-specific

Changelog

UnreleasedCopy Anchor Link

1.7.0-rc.18 — 2026-10-04Copy Anchor Link

SecurityCopy Anchor Link

ChangedCopy Anchor Link

FixedCopy Anchor Link

1.7.0-rc.17 — 2026-10-02Copy Anchor Link

SecurityCopy Anchor Link

FixedCopy Anchor Link

1.7.0-rc.16 — 2026-09-15Copy Anchor Link

FixedCopy Anchor Link

1.7.0-rc.15 — 2026-09-10Copy Anchor Link

FixedCopy Anchor Link

1.7.0-rc.14 — 2026-09-08Copy Anchor Link

FixedCopy Anchor Link

1.7.0-rc.13 — 2026-09-08Copy Anchor Link

FixedCopy Anchor Link

1.7.0-rc.12 — 2026-09-06Copy Anchor Link

FixedCopy Anchor Link

DocumentationCopy Anchor Link

1.7.0-rc.11 — 2026-09-05Copy Anchor Link

FixedCopy Anchor Link

DocumentationCopy Anchor Link

1.7.0-rc.10 — 2026-09-04Copy Anchor Link

FixedCopy Anchor Link

1.7.0-rc.9 — 2026-09-03Copy Anchor Link

SecurityCopy Anchor Link

FixedCopy Anchor Link

DocumentationCopy Anchor Link

1.7.0-rc.8 — 2026-09-03Copy Anchor Link

FixedCopy Anchor Link

1.7.0-rc.7 — 2026-08-29Copy Anchor Link

SecurityCopy Anchor Link

FixedCopy Anchor Link

PerformanceCopy Anchor Link

DocumentationCopy Anchor Link

ChoreCopy Anchor Link

1.7.0-rc.6 — 2026-08-29Copy Anchor Link

AddedCopy Anchor Link

SecurityCopy Anchor Link

FixedCopy Anchor Link

1.7.0-rc.5 — 2026-08-27Copy Anchor Link

SecurityCopy Anchor Link

AccessibilityCopy Anchor Link

FixedCopy Anchor Link

DocumentationCopy Anchor Link

ChoreCopy Anchor Link

1.7.0-rc.4 — 2026-08-26Copy Anchor Link

SecurityCopy Anchor Link

FixedCopy Anchor Link

1.7.0-rc.3 — 2026-08-23Copy Anchor Link

AddedCopy Anchor Link

ChangedCopy Anchor Link

FixedCopy Anchor Link

1.7.0-rc.2 — 2026-08-20Copy Anchor Link

AddedCopy Anchor Link

ChangedCopy Anchor Link

FixedCopy Anchor Link

FixedCopy Anchor Link

RemovedCopy Anchor Link

RemovedCopy Anchor Link

SecurityCopy Anchor Link

DocumentationCopy Anchor Link

1.7.0-rc.1 — 2026-08-14Copy Anchor Link

AddedCopy Anchor Link

ChangedCopy Anchor Link

RemovedCopy Anchor Link

SecurityCopy Anchor Link

FixedCopy Anchor Link

1.6.0 — 2026-08-11Copy Anchor Link

AddedCopy Anchor Link

ChangedCopy Anchor Link

DeprecatedCopy Anchor Link

RemovedCopy Anchor Link

FixedCopy Anchor Link

SecurityCopy Anchor Link

Upgrade NotesCopy Anchor Link

DocumentationCopy Anchor Link

1.6.0-rc.13 — 2026-08-08Copy Anchor Link

SecurityCopy Anchor Link

Unreleased

1.7.0-rc.18 — 2026-10-04

Security

Changed

Fixed

1.7.0-rc.17 — 2026-10-02

Security

Fixed

1.7.0-rc.16 — 2026-09-15

Fixed

1.7.0-rc.15 — 2026-09-10

Fixed

1.7.0-rc.14 — 2026-09-08

Fixed

1.7.0-rc.13 — 2026-09-08

Fixed

1.7.0-rc.12 — 2026-09-06

Fixed

Documentation

1.7.0-rc.11 — 2026-09-05

Fixed

Documentation

1.7.0-rc.10 — 2026-09-04

Fixed

1.7.0-rc.9 — 2026-09-03

Security

Fixed

Documentation

1.7.0-rc.8 — 2026-09-03

Fixed

1.7.0-rc.7 — 2026-08-29

Security

Fixed

Performance

Documentation

Chore

1.7.0-rc.6 — 2026-08-29

Added

Security

Fixed

1.7.0-rc.5 — 2026-08-27

Security

Accessibility

Fixed

Documentation

Chore

1.7.0-rc.4 — 2026-08-26

Security

Fixed

1.7.0-rc.3 — 2026-08-23

Added

Changed

Fixed

1.7.0-rc.2 — 2026-08-20

Added

Changed

Fixed

Fixed

Removed

Removed

Security

Documentation

1.7.0-rc.1 — 2026-08-14

Added

Changed

Removed

Security

Fixed

1.6.0 — 2026-08-11

Added

Changed

Deprecated

Removed

Fixed

Security

Upgrade Notes

Documentation

1.6.0-rc.13 — 2026-08-08

Security

Fixed