Skip to content

fix(OperationsCenter): stop Rule 8 retries on backend-capacity planner failures - #471

Merged
ProtocolWarden merged 1 commit into
mainfrom
oc-watchdog/20260715-0027-backend-capacity-loop
Jul 15, 2026
Merged

fix(OperationsCenter): stop Rule 8 retries on backend-capacity planner failures#471
ProtocolWarden merged 1 commit into
mainfrom
oc-watchdog/20260715-0027-backend-capacity-loop

Conversation

@ProtocolWarden

Copy link
Copy Markdown
Owner

Summary

  • board_worker now labels tasks blocked-reason: backend-capacity when a blocked task fails with the exact non-JSON + session limit or error backend signature.
  • board_unblock Rule 8 (CLEAN_BLOCKED_RETRY) now excludes tasks carrying that label, so backend-capacity failures no longer get recycled into an immediate Blocked -> Backlog -> Ready for AI -> Blocked loop.
  • Root cause: Rule 8 only excluded explicit dependency/signal/exit-code/policy-gate labels; a bare backend-capacity failure had no structured label, so it was treated as a generic clean retry and re-dispatched into the same failure repeatedly (closed-loop stagnation observed across the 2026-07-15 watchdog cycles).

Test plan

  • pytest tests/unit/entrypoints/board_worker/test_outcomes_cov.py tests/unit/entrypoints/maintenance/test_board_unblock.py tests/unit/entrypoints/maintenance/test_board_unblock_cov.py -q → 202 passed
  • pytest tests/unit/er000_phase0_golden/ -q → 15 passed

@ProtocolWarden
ProtocolWarden merged commit 1d1bc38 into main Jul 15, 2026
25 checks passed
@ProtocolWarden
ProtocolWarden deleted the oc-watchdog/20260715-0027-backend-capacity-loop branch July 15, 2026 05:54
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant