release: forjar 1.33.0-rc.1 (#647 + #648) - #650
Merged
Merged
Conversation
One wall-clock sample measured the scheduler: 84ms at nice 19 under the parallel lib suite failed the 1.33.0-rc.1 clean-room, the only red out of 13566. A query that stops using the FTS index is slow on every run, so the minimum still catches the regression the test exists for. Refs #647 Pmat-Ticket: PMAT-642 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… and a stack is drift (fixes #648) an autofs trigger. mount.cifs fired systemd's automount, systemd mounted its generated unit's OLD options (gid=1000) underneath, and forjar's mount landed on top. check read findmnt | tail -1, the top only, and said converged. - fstab is written FIRST; an automount entry then gets systemctl daemon-reload BEFORE the detach, and is mounted by touching the path, never by forjar's own mount. - The detach is a bounded loop over every filesystem stacked at the target, a plain umount, busy failing loudly (lazy only on the wrong-source branch, as before). - check: more than one filesystem mounted at the target is drift. tests/falsification_648_automount_stack.rs models the kernel stack and systemd's unit and runs the generated scripts: 4 of 6 go red with mount.rs reverted to main. Pmat-Ticket: PMAT-648 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An rc tag is a fleet-deploy candidate, not a release. bench.yml and coverage.yml each hold a bare-metal intel runner for ~50 minutes on a v* tag; the final v1.33.0 tag still runs both. Refs #647 Pmat-Ticket: PMAT-642 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…taches (#648) Quorum round 1 on #650 found that the wrong-source branch passed lazy=true to remount() regardless of automount, so a busy stack under an automount fell back to `umount -l || true` and detached live writers while systemd re-mounted underneath. The lazy fallback is now kept for a plain mount only. a_busy_wrong_source_under_automount_fails_loudly_and_never_lazy_detaches goes red with the fix reverted (umount -l called) and green with it. Refs #648 Pmat-Ticket: PMAT-648 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This reverts commit 02acddf. The rc-tag trigger change is outside the scope of #647/#648 (quorum round 1 on #650); it lands separately on its own ticket if it is wanted. The rc tag runs bench and coverage like any v* tag. Refs #647 Pmat-Ticket: PMAT-648 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
noahgift
changed the title
noahgift
enabled auto-merge (squash)
September 28, 2026 08:30
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
1 participant
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.
Release PR for the fleet-deploy candidate v1.33.0-rc.1 (closes #651).
x-systemd.automountremount no longer stacks a stale mount, and a stack is drift. The check counts every non-autofs filesystem at the target. For an automount entry, apply writes fstab, runssystemctl daemon-reloadbefore detaching, detaches the whole stack with a plainumount, and lets systemd mount on a path touch. A busy stack fails loudly and is never lazily detached. Plain mounts are unchanged.query_latency_under_50msasserts the best of 5 samples (test-only).Cargo.toml/Cargo.lock: 1.33.0-rc.1.Plan: merge normally. After main CI passes on the merge sha,
v1.33.0-rc.1is tagged on that main sha. No crates.io publish.Quorum:
.quorum/release-v1.33.0-rc.1.json, 3 rounds, 3/3 PASS at d1c5b54. Round 1 found a lazy detach on the wrong-source automount path (fixed in d07790c, with a test that goes red without the fix) and an out-of-scope workflow change (reverted in 2ab9f59). Evidence is in.quorum/evidence/PMAT-651-*.md.Closes #651
Fixes #648
Fixes #647
🤖 Generated with Claude Code