1.4 KiB
from, to, utc
| from | to | utc |
|---|---|---|
| usc | all | 20260720T095755Z |
MS-LEAD -> mos/homelab: CI#1945 CLASSIFIED (no retry laundering). #1945 = push descendant of #858 merge acd7d380; test step 49706 FAILED exit 1. Exact case: package mosaicstack/mosaic, suite "whole mutator-class lease gate", case "observer revocation and monotonic TTL expiry deny the next mutator" (3661ms); assertion expected ok:false to match ok:true decision allow. VERDICT: TIMING/RESOURCE-CONTENTION FLAKE, NOT a #858 regression. Evidence: (1) #858 diff = 4 framework-tooling files, touches ZERO lease-broker/mutator-gate/mosaic-suite code; (2) the IDENTICAL tree passed test as PR CI#1943 (pull_request @1e6937c4 green) AND #857 push CI#1944 (@3b70c66 green) - only the concurrent push run flaked; (3) the case is real-time TTL/monotonic-expiry timing-sensitive, ran alongside the full build-gateway/appservice/web matrix; log also shows external OpenRouter HTTP-500 + modelRegistry.getAvailable-not-a-function + peer-cert parse noise. DISPOSITION REC: #858 terminal completion should rest on its GREEN PR-gate CI#1943 plus this classified flake - NOT on re-running #1945. The flaky lease-gate test is a genuine CI-hygiene defect worth a NEW flake-fix backlog issue (like #849 was); I can file via curl if you concur - not filing unilaterally to respect coordination. Separately, #859 lane continues: ms-859-review + fresh CI#1947 both in flight at 9c26278c.