Skip to content

test(v0.49): retain two-case backlog progress and drain latency evidence - #19

Merged
grove merged 3 commits into
issue/11-explicit-escalation-timefrom
test/v049-backlog-observation-timeline
Oct 21, 2026
Merged

grove merged 3 commits into
issue/11-explicit-escalation-timefrom
test/v049-backlog-observation-timeline

Conversation

@grove

@grove grove commented Oct 21, 2026 •

Copy link
Copy Markdown
Contributor

CI repair — 2026-10-21

The initial exact-head pinned producer run #39673424319 completed the original full joint-stack audit, one-case outage test and new two-case managed-outage test successfully. The two-case test printed its final PASS after asserting immutable vectors, monotone case-key progress and no replay. The step then failed in the shell wrapper only: unexpected EOF while looking for matching \', caused by an unmatched single quote at the end of the appended inline python3 -c assertion. No runtime or numerical gate failed.

Commit feb85b2 closes that quote. The existing complete runtime and final evidence assertions remain unchanged. Fresh exact-head CI is pending; do not merge until the current-head pinned test and all other required checks are green.

Focused backlog observation and drain-budget evidence

Based on the exact green React feature PR #14 head 1d44dfda473591bea19c88a0a3f4078560c553c1. Extends the already-qualified real two-case MDM outage in audit.py --multi-outage with explicit monotone backlog progress observations, without changing production runtime, worker scheduling, SQL, authorization, or pinned MDM/pg-trickle images.

  • Capture every sampled managed status during recovery as elapsed monotonic time, completed case-key set, count, worker state and public pending/processed indicators. Evidence is stored in the JSON artifact.
  • Fail closed on disappearing applied receipts, out-of-scope case identities, duplicate receipts, missing progress or incomplete drain; assert both distinct cases complete inside the existing 150-second bounded recovery wait.
  • Enforce the recorded timeline and drain elapsed-time fields in the original pinned compatibility workflow. Preserve the full existing audit, one-case and two-case outage tests, complete immutable state vectors, second restart/no-replay, and source/binary hashing.

Limitations: The 150-second wait here is a small focused two-case runtime test, not the signed T49.12 300-case p95/maximum target or an approved operational bound. Sampling cannot by itself prove actual per-cycle admission count or mixed-backlog fairness. Those need a dedicated instrumentation/acceptance profile; likewise clone/restore, 60/min throughput and signed owner envelope remain open.

Draft, pending current-head CI. Merge only into issue/11-explicit-escalation-time after clean exact-head checks, never main. Keep PR #14 and issues #11/#12/#8 open.

@grove
grove marked this pull request as ready for review October 21, 2026 21:48
@grove
grove merged commit d70a73d into issue/11-explicit-escalation-time Oct 21, 2026
7 checks passed
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