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
🐛 [reviewer] PR stays at 070 Fixing for hours after the fixer exhausted its attempts #15744
Hive-authored PR #15308 (XXL, adopted from #15226 for #15198) sat at stage/070-fixing for 13h with the tracker reading "Fixing · @hivecommons-hive[bot]" — while nobody was fixing it.
The PR is also CONFLICTING with v5. The fix dispatcher excludes it (fix-attempts-exhausted, cap = 6 re-engagements), automerge skips it, yet the stage engine keeps it on 070 and the dashboard says the hive is on it. Worse, stage/stalled was removed:
— a stale answered decision on a head that has not moved re-cleared the stall signal.
Expected
When the fix dispatcher excludes a PR with fix-attempts-exhausted, the stage engine moves it from 070 to 085 Needs decision with caption fixer gave up after N attempts — <reviewer summary first line> and posts one decision comment per head (options: re-author from scratch, hand to a person, close as not planned). Tracker lead action: Re-author (decide:re-author); "Close as not planned" lives in the Decide… submenu. Write-back gating respected.
The 070 caption never claims the hive is fixing when no fix job is dispatched/active: it says why the fixer is excluded (exhausted / conflicting / no fixer).
A stale decision-answered must not clear stage/stalled / flip the stage when the head has not moved and the fixer is excluded.
The review fix dispatch excluded WARN logs once per (pr, head), not every cycle.
Activity