[Danathar]

Summary

Follow-up to the triage on #14466. The hive parked three adopted contributor PRs on the projectbluefin hive at stage/085 as adoption-stalled:hive-signoff-attribution, right after its own fix commit:

• projectbluefin/utah-packages#660
• projectbluefin/chairlift#638
• projectbluefin/testsuite#1028

The park itself is designed (#13800, #13803). Two things around it were wrong, and this changes only those:

1. The escalation was logged at WARN. The slog feed observes WARN and above, so three designed parks inside ten minutes became an unclassified_error_repeated fleet self-report ([selfreport] unclassified_error_repeated: stage labels: adopted PR sits with nothing the hive can push; escalating to nee… #14466). It is now logged at INFO, as the triage suggested.
2. The notes said the Sign-off survives the squash gate "stays red". That check is .github/workflows/dco-squash-attribution.yml, which only some repositories carry. None of the three repos above runs it; the hive's own commit check is what held them. The notes now say the hive will not squash-merge the head, and name the check only when the head's required or reported checks include it.

The log-level line is byte-identical to #14480, which the hive's scanner opened against v5. This PR carries it to v6 with a regression test, so the v5→v6 forward-merge meets the same line on both sides.

Cause

At 392df48:

• automerge_contributor_adopt.go:1179 logs the escalation with c.warn. slogfeed.go:31 observes WARN and above.
• automerge_contributor_adopt.go:455 (the adoption stall note) and automerge_merge_commit.go:161 (the maintainer sign-off note) always say the gate stays red. The hold comes from the stage pass's own commit check (automerge_stage_dco.go:67), on every managed repository.

Change

• Log level: the "adopted PR sits with nothing the hive can push" line is c.info. The escalation, its note and its stage are unchanged.
• Gate detection: noteStageDCOGate already reads the head's required and reported check names for the dco context. It now also records SquashSignoffGate when one of them is Sign-off survives the squash, for the same head as DCOGateHead. The tracker copies it with the DCO gate, under the same head guard.
• Note wording: both notes say "…so the hive will not squash-merge this head". They add "(the Sign-off survives the squash check stays red)" only when that check was seen on this head (squashSignoffGateNote). The maintainer asks and the merge-commit route are unchanged.
• Docs: src/docs/labels-and-control-signals.md said the required check "is red there" for every such head. It now says the hive will not squash-merge the head, that the check is red too only where a repository runs the workflow, and that the notes name the check only then.

Testing

From src/ under /var/tmp, -count=1:

• go test ./pkg/github/... ./pkg/stageengine/... ./pkg/stagelabels/... ./pkg/fleetreport/... ./pkg/review/...: all ok.
• go build ./pkg/... ./cmd/... succeeds; go vet ./pkg/github/automerge ./pkg/stageengine/... and gofmt are clean.
• check-docs-links.py and check-docs-citations.py pass on src/docs; check-changelog-fragment.sh v6 passes.

New and changed tests:

Test	Covers
TestSignoffStallNoteNamesSquashGateOnlyWhereItRuns	The adoption stall note on the #1028 shape, without and with the check on the head. Goes through noteStageDCOGate and the tracker.
TestSignoffAttributionDecisionNoteNamesSquashGateOnlyWhereItRuns	The same for the maintainer sign-off note on a non-adopted, hive-caused head (the testsuite#991 fixture).
TestSquashSignoffGateNoteIsPerHead	A check seen on an earlier head is not named.
TestAdoptionStallEscalationLogsAtInfo	The escalation logs at INFO, and the real slogfeed handler observes nothing for it.
TestAdoptedInPlaceSquashUnsafeHiveCommitStallsWithoutSupersede	Existing. Its doc says the gate rejects the head, so it now notes the check on that head; it still asserts the note names it.

Mutation checks

Each part was broken in turn and restored from a backup:

• Back to WARN: the log test fails.
• Always name the check: all three wording tests fail.
• No detection in noteStageDCOGate: the existing stall test and both wording tests fail.
• No copy in the tracker merge: the same three fail.
• No head guard: the per-head test fails.

Not verified

• No live hive. The three projectbluefin PRs were parked by fcf0e78; their notes are not rewritten (one note per head).
• The stall itself is unchanged. A separate question about when it fires (it removes stage/074, which stops the review fixer on the hive's own head) is raised on [selfreport] unclassified_error_repeated: stage labels: adopted PR sits with nothing the hive can push; escalating to nee… #14466 as a proposal, not changed here.

Contributor checklist

• Changelog: changelog.d/fixed-signoff-stall-note-and-log-level.md.
• Targets v6.
• The title uses the emoji convention.
• Commits include DCO sign-off.
• Docs updated (src/docs/labels-and-control-signals.md).
• No secrets, credentials or local runtime state are committed.

🤖 Generated with Claude Code

https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1

[kubestellar-prow]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign danathar for approval. For more information see the Code Review Process.

The full list of commands accepted by this bot can be found here.

Details

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[hivecommons-hive]

@Danathar this PR is back in your court (stage/072-returned-to-contributor). The hive will not merge it until the head changes.

Failing checks:

• check-failure

Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path.

Your relay session received this as a fix-pr task; your current task is parked (claim kept alive) until this PR is green.

[hivecommons-hive]

Looks correct to me. Two notes for the maintainer.

The red checks are not this PR's. test (agent 4/5) (job 121943222780) fails TestPollTmuxOutputForAgent_WiresThrashBreaker 3/3 at src/pkg/agent/thrash_poller_wiring_test.go:110; coverage and test are the roll-ups of that shard. This PR touches nothing under pkg/agent. That is the v6 half of incident #14406, and #14494 forward-ports the #14415 fix. Once #14494 lands, a re-run here should be green.

The change itself, read against v6:

• squashSignoffGateNote (automerge_merge_commit.go:157-163) returns the check mention only when Detail.SquashSignoffGate is set and DCOGateHead is the observed head. The flag is set in noteStageDCOGate (automerge_stage_labels.go:563-575) from the same required/observed sets as DCOGate, and StageTracker.drain (:296) only copies it inside the existing DCOGateHead == o.HeadSHA guard, so the per-head semantics hold on both paths and TestSquashSignoffGateNoteIsPerHead pins them.
• c.info exists on v6 (automerge_sweep.go:3213) and slogfeed.minLevel = slog.LevelWarn (slogfeed.go:31), so the INFO line genuinely drops out of the self-report feed; TestAdoptionStallEscalationLogsAtInfo drives the real handler rather than mocking it.
• Both call sites that carried the old "gate stays red" wording (adoptionBlockerText at automerge_contributor_adopt.go:455 and SignoffAttributionDecisionComment at automerge_merge_commit.go:182) are updated; no other non-test occurrence remains on the head.
• The log-level change at automerge_contributor_adopt.go:1179 is the same one-line edit as #14480 on v5, as the body says, so the forward-merge meets the same line on both sides.

— hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1

[hivecommons-hive]

hive/shared-ci: the test (agent 4/5) failure here (TestPollTmuxOutputForAgent_WiresThrashBreaker, 3/3 on ubuntu-26.04) is not caused by this PR's diff. It is a v6-line shared incident: the #14415 fix landed on v5 only and has not been forward-ported to v6 (src/pkg/agent/thrash_poller_wiring_test.go). Tracked in #14506; a forward-port PR is being opened. Once it merges, update this branch from v6 and the slice should go green — no change needed in this PR.

---

🐝 Hive Agent: scanner | Instance: hosted-available-oke-11-placeholder-r05x | SHA: b05e7ba6f

— hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88

[hivecommons-hive]

stage 060-reviewed → 050-review-pending: shared-ci-incident:14506

[github-actions]

Open PR overlap check

Base branch: v6

Conflicts are checked with every PR merged onto the current tip of v6, so a PR that only needs a rebase is not reported as conflicting with its siblings.

No conflicting open PRs or same-file overlaps were found.

[hivecommons-hive]

Shared-CI incident #14506 is resolved: the test (agent 4/5) fix landed on v6 via #14494. This PR only needs its branch updated from v6 (no code change) for the shard to go green; the merge sweep will pick it up once CI passes.

---

🐝 Hive Agent: scanner | Instance: hosted-available-oke-11-placeholder-r05x | SHA: unknown

— hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88

[hivecommons-hive]

hive/awaiting-ci: branch synced with v6 (picks up #14494, which fixed the TestPollTmuxOutputForAgent_WiresThrashBreaker failure on test (agent 4/5)). CI pending — sweep will merge when green.

---

🐝 Hive Agent: scanner | Instance: hosted-available-oke-11-placeholder-r05x | SHA: unknown

— hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88

[hivecommons-hive]

@Danathar this PR is back in your court (stage/072-returned-to-contributor). The hive will not merge it until the head changes.

Failing checks:

• check-failure

Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path.

Your relay session received this as a fix-pr task; your current task is parked (claim kept alive) until this PR is green.

[hivecommons-hive]

Read automerge_merge_commit.go, automerge_contributor_adopt.go, automerge_stage_labels.go and prstage/types.go at 101417b; this looks correct to me.

• SquashSignoffGate is written in noteStageDCOGate (automerge_stage_labels.go:573) together with DCOGateHead, and drain() (:294-297) only copies it under the same head guard as DCOGate, so a stale value cannot carry across heads. squashSignoffGateNote (automerge_merge_commit.go:157-163) repeats that guard. Both note sites now use it; no non-test note text still hardcodes the gate name.
• warn → info at automerge_contributor_adopt.go:1179 is a log-level change only.

CI: test (agent 1/5) is red on TestSendKick_RestartsCrashedCLIThenDelivers and TestRestart_CancelsPendingAsyncKick in pkg/agent, which this diff does not touch; coverage fails only because that shard failed ("Coverage requires successful test shards … agent=failure"). That looks like an unrelated flake rather than something for this PR to fix. go build ./pkg/github/automerge/ ./pkg/stageengine/... passes locally.

— hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88

[hivecommons-hive]

Read the diff against the v6 tree. Looks correct to me.

• .github/workflows/dco-squash-attribution.yml:34 — job name is exactly Sign-off survives the squash, matching squashSignoffCheck (automerge_merge_commit.go:147); isSquashSignoffCheck compares case-insensitively.
• automerge_stage_labels.go:557-573 — noteStageDCOGate is called unconditionally from the sweep for every PR (automerge_sweep.go:2961) over both required and st.Observed, so SquashSignoffGate is populated wherever DCOGate is; drain() (line 294-296) copies it only under the same head match, so a stale head reads as "not present" and the note falls back to the plain wording.
• escapeStalledAdoption WARN→INFO (automerge_contributor_adopt.go:1179) matches the v5 line in #14480, as the body says.
• Builds clean at 101417b.

The red test (agent 1/5) / coverage on run 40478131307 are pkg/agent tmux restart tests (kick_cancel_test.go:299, kick_restart_recovery_test.go:132), unrelated to this diff.

— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88

[hivecommons-hive]

hive/check-report (head 101417b): the test (agent 1/5) failure (TestSendKick_RestartsCrashedCLIThenDelivers, TestRestart_CancelsPendingAsyncKick) was transient — this PR does not touch src/pkg/agent, v6 is green on every agent shard, and the re-run on the same head passed (job 122105865852). Every check is now green at 101417b. Since the PR was returned to contributor on that red run, it needs a new head to re-enter the merge path: an empty signed commit (git commit -s --allow-empty -m 'ci: re-run') is enough; no code change required.

---

🐝 Hive Agent: scanner | Instance: hosted-available-oke-11-placeholder-r05x | SHA: unknown

— hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88

[hivecommons-hive]

Evidence bundle `#14481101417b` sha256:a4e94d22d56d9ee5acc1c9d30fe2520a9dc06151957ea4c4fc9d094772060706 — download from the dashboard or `hivectl review evidence #14481`.

[hivecommons-hive]

merged by Hive trusted-author auto-merge: author @Danathar holds role owner and GitHub permission maintain on repo; CI green at 101417b.

[hivecommons-hive]

Reviewed (correctness) — no blocking findings at 101417b. Two advisory notes:

1. The check can be missed on a repo that runs it (conservative miss, but permanent per head). SquashSignoffGate is only set from commitGreenForPRWith (src/pkg/github/automerge/automerge_sweep.go:2961), and the tracker clears t.details every drain (src/pkg/github/automerge/automerge_stage_labels.go:304). If the cycle in which escapeStalledAdoption first fires bails before CI evaluation (behind/conflicting/missing-head paths at automerge_sweep.go:1690-1703), the one-per-head note at automerge_contributor_adopt.go:1183 is written without the mention and is never rewritten. Same cycle-scoped behaviour as the pre-existing DCOGate, and omitting the mention is the safe direction, so advisory only — but the body's "name the check only when the head's … checks include it" is really "when they were observed in the escalating cycle".

2. The INFO demotion covers every stall reason. c.info at automerge_contributor_adopt.go:1179 runs for whatever adoptionStalledReason returns, not only hive-signoff-attribution. That matches #14480 on v5 and the PR body's description of "the line", so no change requested; just noting the scope is broader than the title suggests.

Verified: Observed holds check-run names (src/pkg/github/commit_ci.go:102) and the workflow's job name is Sign-off survives the squash (.github/workflows/dco-squash-attribution.yml:34), so isSquashSignoffCheck matches in production. The facts round-trip (src/pkg/stageengine/derive_pr.go:151-157) drops SquashSignoffGate, but observationsFrom is test-only, so notes are unaffected.

— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88

[hivecommons-hive]

Read the diff against the tree at 101417b, including the callers. Looks correct to me.

• noteStageDCOGate is fed from automerge_sweep.go:2961 with st.Observed, which commit_ci.go:437 fills with every status-context/check-run name on the head — so isSquashSignoffCheck will see Sign-off survives the squash exactly when the workflow (.github/workflows/dco-squash-attribution.yml:34, name: Sign-off survives the squash) actually reported on this head.
• StageTracker.drain (automerge_stage_labels.go:293-296) copies SquashSignoffGate under the existing DCOGateHead == o.HeadSHA guard, and squashSignoffGateNote re-checks the head, so a gate seen on an earlier head is not named on the new one.
• Sweep order holds: SweepContributorLaneAutoMerges (CI evaluation) runs before ApplyStageLabels at automerge_sweep.go:1533, matching the order the new tests set up.
• go build ./pkg/github/automerge/ ./pkg/stageengine/... is clean at this head; CI is green (Sign-off survives the squash passed on the latest run; the three CANCELLED entries are superseded runs).

Body matches the diff: the only behaviour changes are the WARN→INFO level at automerge_contributor_adopt.go:1179 and the two note strings.

— hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88