Repository navigation
chore(security): record the 2026-11-16 recheck of the zlib1g findings in the pinned Postgres image - #275
Merged
Conversation
… in the pinned Postgres image The scheduled image audit failed on 9 and 10 November because the official postgres and redis Debian images ship zlib1g 1:1.3.dfsg+really1.3.1-1+b1 with CVE-2026-76844 and CVE-2026-85091, which Debian fixed in +deb13u1. #266 ignored both findings until 2026-11-20, #270 moved the panel's Redis to redis:8.10.3-alpine and #271 moved Postgres to 18.7; the audit has been green since 2026-11-13. Rechecked on 2026-11-16: postgres:18.7 (index sha256:2ea8630d..., built 2026-11-12) and redis:8.10.3 (Debian, built 2026-11-05) still carry the old zlib1g; there is no postgres 18.8 or redis 8.10.4. redis:8.10.3-alpine has zlib 1.3.2-r1 and scans clean. A scan of postgres:18.7 with the workflow's pinned Trivy and the gosu skip passes, the two zlib1g findings ignored. So the pins stay, and the ignore entries keep their 2026-11-20 expiry, after which the audit goes red and the image is checked again. - .trivyignore.yaml: the zlib1g statements say the 2026-11-12 rebuild still ships the old version and when that was checked. - image-audit.yml: the gosu comment names the image it is about now (postgres 18.7, rebuilt 2026-11-12, gosu 1.19 on Go 1.24.6). - docs/agenthits-install.md, docs/server-components-update.md: the installer defaults read postgres:18.7 and traefik:v3.7.19 as in install-agenthits.sh, not 18.6 and v3.7.17. - CLAUDE.md and docs/server-components-update.md: Postgres and Redis are pinned by tag only. A same-tag rebuild reaches new installs, which pull the tag, but not a panel updated with the installer, which replaces dokploy-postgres and dokploy-redis only when the tag changes, and pins-update.sh follows new tags, not rebuilds. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…pdate crash test Since #211 a restart loop fails as soon as one slot has three failed tasks. "fails when an update changes nothing and the service keeps crashing" still expected the limit's message only, so it failed whenever the runner saw the third crash before the limit: pr-check (docker-1) on #275 did. Both messages are right for this case; the test now takes either, and still requires the task's own `non-zero exit (3)`. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
The scheduled image audit failed on 9 and 10 November: the official
postgresandredisDebian images shipzlib1g1:1.3.dfsg+really1.3.1-1+b1 with CVE-2026-76844 and CVE-2026-85091, which Debian fixed in+deb13u1. #266 ignored both findings until 2026-11-20, #270 moved the panel's Redis toredis:8.10.3-alpine, and #271 moved Postgres to 18.7. The audit has been green since 2026-11-13. Four days before the ignore entries expire, I checked whether upstream has a fixed image to move to.Upstream on 2026-11-16 (checked from the build VM):
postgres:18.7(=18,18.7-trixie)sha256:2ea8630df176…zlib1g1:1.3.dfsg+really1.3.1-1+b1redis:8.10.3(Debian, =8.10,8)sha256:1f88291b4c57…zlib1g1:1.3.dfsg+really1.3.1-1+b1redis:8.10.3-alpine(pinned)sha256:7958cfbb9d15…zlib1.3.2-r1postgres:18.8andredis:8.10.4do not exist. No Debian image has the fix yet, so there is nothing to move the pins to. The ignore entries keep their 2026-11-20 expiry, so the audit goes red then and the image gets checked again.What changes
No pin and no expiry date change. This PR updates text only:
.trivyignore.yaml: the zlib1g statements say the 2026-11-12 rebuild of 18.7 still ships the old version, and when that was checked.image-audit.yml: the gosu skip comment names the image it is about now: postgres 18.7, rebuilt 2026-11-12, still on gosu 1.19 built with Go 1.24.6. This is a comment only, and the sync test's "names no pinned version" check still passes.docs/agenthits-install.md,docs/server-components-update.md: the installer defaults now readpostgres:18.7andtraefik:v3.7.19, as ininstall-agenthits.sh. They said 18.6 and v3.7.17.CLAUDE.mdanddocs/server-components-update.md: one sentence each saying Postgres and Redis are pinned by tag only:dokploy-postgres/dokploy-redisonly when the tag changes, so a running panel keeps its old build.pins-update.shfollows new tags, not rebuilds.Checks
On the build VM, in a fresh clone of
AgentHits-Devwith this patch:.github/scripts/trivy-image-scan.shand the workflow's pinnedaquasec/trivy:0.76.1@sha256:790fd66d…,--image-src remote --platform linux/amd64. All three exit 0:postgres:18.7with the gosu skip: 0 critical/high with a fix, the 2 zlib1g findings ignored.redis:8.10.3-alpine: 0 findings.redis:8.10.3(Debian): 0 with a fix, the 2 zlib1g findings ignored.vitest run __test__/server/component-versions-sync.test.ts __test__/settings/install-agenthits-update.test.ts: 2 files, 103 tests passed.biome check --writeon the changed files: no changes to the YAML orCLAUDE.md. Biome wants to pad every Markdown table in both docs files, and the base files already fail that check. Those rewrites are left out to keep the diff small.Not covered: the 2026-11-20 run itself, and the 26 Go stdlib findings that gosu 1.19 brings back on 2026-11-27.
🤖 Generated with Claude Code