You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Reduce HIGH findings on clamav/clamav latest + latest_base (1 HIGH each, 0 CRITICAL — pcre2 CVE-2026-103111) #111
Reduce HIGH findings on clamav/clamav (1 HIGH × 2 services, 0 CRITICAL)
Status
Open. Surfaced 2026-10-04 00:24 UTC by the weekly-pattern CI scan when the unpinned:latest/:latest_base tags drifted (September scans showed 0 HIGH). Both services are now digest-pinned in docker-compose.yaml (moved 1fdfd24c.../5a7c486f... -> 7659dcb0.../4bd75811...) so the baseline can no longer move silently.
CVEs (Trivy 0.74.0, DB 2026-10-04, HIGH+CRITICAL, ignore-unfixed)
CVE-2026-103111 [HIGH] pcre2@10.48-r0 -> 10.49-r0 (out-of-bounds write via crafted regular expression) — present in BOTH digests. Status: fixed in the Alpine package feed; awaiting ClamAV image rebuild.
Why NOT in .trivyignore
Deliberate: unlike #83/#84/#87 (no upstream rebuild in flight), ClamAV publishes updated images frequently and the Alpine fix already exists. Suppressing would blind the re-pin signal; the finding should stay visible until the next digest bump clears it.
Re-pin both services to the new digests, re-scan, and update docs/container_security_scan.md + tests/test_container_security_scan.py::test_compose_clamav_digest_pinned.
Acceptance criteria
Both clamav services re-pinned to digests with 0 HIGH / 0 CRITICAL, or
Pins + guard test + docs shipped in the commit following this issue (#111). CI scan (push-triggered) will now scan the pinned digests directly; expect the clamav/clamav-base jobs to report exactly 1 HIGH (CVE-2026-103111) each until upstream rebuilds with pcre2 >= 10.49-r0, then a Renovate/manual re-pin clears it.
Re-pin signal restored: the CI matrix now scans the floatingclamav/clamav:latest + :latest_base tags (merged PR #113) while compose keeps the pinned 2026-10-04 digests as the deployment baseline. Verified 2026-10-10: registry digests are still byte-identical to our pins — upstream fix pending, as expected. Behavior from here: weekly/push scans will report the 1 HIGH (pcre2 CVE-2026-103111) per clamav job on the floating tags; the run where those jobs drop to 0 is the re-pin trigger. Then: bump compose digests + workflow, refresh docs rows 4-5, and this issue can close. 6-day recheck of alert convergence: Security tab now shows exactly one open Trivy HIGH — alert #494 (this CVE) — 43 stale Sept-13 orphans were dismissed as mitigated with cross-links.
Digest re-check (2026-10-10, live docker manifest inspect): registry clamav/clamav:latest → sha256:7659dcb0db47… and :latest_base → sha256:4bd758114dbe… — both unchanged since the 2026-10-04 pins. Upstream has not yet published the pcre2 fix (pending 10.49-r0), so the wait stays correct and compose pins hold.
Post-merge CI corroborates: the floating-:latest matrix jobs (clamav, clamav-base) in run 38105001331 on main still report the single 1 HIGH pcre2 finding each — same as the pinned deployments, no new movement. Re-pin trigger unchanged: when the floating scans drop to 0 findings, verify the candidate digest, then bump compose + matrix + guard test + docs together.
Activity
chore(security): digest-pin clamav services, track pcre2 finding (#111)
JJediny commentedon Oct 4, 2026
Pins + guard test + docs shipped in the commit following this issue (#111). CI scan (push-triggered) will now scan the pinned digests directly; expect the clamav/clamav-base jobs to report exactly 1 HIGH (CVE-2026-103111) each until upstream rebuilds with pcre2 >= 10.49-r0, then a Renovate/manual re-pin clears it.
chore(security): digest-pin the grype scanner image; scan the deploye…
fix(ci): scan floating clamav tags as the #111 drift monitor (#113)
JJediny commentedon Oct 11, 2026
Re-pin signal restored: the CI matrix now scans the floating
clamav/clamav:latest+:latest_basetags (merged PR #113) while compose keeps the pinned 2026-10-04 digests as the deployment baseline. Verified 2026-10-10: registry digests are still byte-identical to our pins — upstream fix pending, as expected. Behavior from here: weekly/push scans will report the 1 HIGH (pcre2 CVE-2026-103111) per clamav job on the floating tags; the run where those jobs drop to 0 is the re-pin trigger. Then: bump compose digests + workflow, refresh docs rows 4-5, and this issue can close. 6-day recheck of alert convergence: Security tab now shows exactly one open Trivy HIGH — alert #494 (this CVE) — 43 stale Sept-13 orphans were dismissed asmitigatedwith cross-links.JJediny commentedon Oct 11, 2026
Digest re-check (2026-10-10, live
docker manifest inspect): registryclamav/clamav:latest→sha256:7659dcb0db47…and:latest_base→sha256:4bd758114dbe…— both unchanged since the 2026-10-04 pins. Upstream has not yet published the pcre2 fix (pending10.49-r0), so the wait stays correct and compose pins hold.Post-merge CI corroborates: the floating-
:latestmatrix jobs (clamav,clamav-base) in run38105001331onmainstill report the single 1 HIGH pcre2 finding each — same as the pinned deployments, no new movement. Re-pin trigger unchanged: when the floating scans drop to 0 findings, verify the candidate digest, then bump compose + matrix + guard test + docs together.