tests: raise Coverage (Debug) timeout for world allocation shards (fix flaky main CI)#3195
tests: raise Coverage (Debug) timeout for world allocation shards (fix flaky main CI)#3195jslee02 wants to merge 1 commit into
Conversation
test_world_raw_malloc timed out at ~3600.1s on the gcov-instrumented Coverage (Debug) job (a sharded allocation gate that already gets a coverage-specific TIMEOUT 3600). It sits right at the limit and load-flakes — the same job also failed on the preceding commit (#3127, an actions/checkout bump), so this is a flaky coverage timeout, not a code regression. Raise the world-shard coverage timeout to 5400s for headroom.
|
Codex Review: Didn't find any major issues. Already looking forward to the next diff. Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
|
Closing in favor of #3196, which fixes this at the root instead of raising the timeout. The timeout bump here was a band-aid. The real cause is that |
1 participant
Fixes the flaky
Coverage (Debug)failure onmain.Symptom:
Coverage (Debug)failed onmain(currently red ondcb5c688260/#3189) — CTest exit 8,test_world_raw_mallocTimeout at 3600.13s.Root cause (not a code regression):
test_world_raw_mallocis one of five sharded world allocation gates that already get a coverage-specificTIMEOUT 3600(the gcov-instrumented build is much slower). It ran ~0.13s over the limit — i.e. it sits right at the boundary and load-flakes. The sameCoverage (Debug)job also failed on the preceding commit #3127 (anactions/checkoutversion bump) and passed on the two before that; neither that bump nor the docs-only #3189 can affect test runtime. So this is a flaky coverage timeout, surfaced by runner load, not anything those commits introduced.Fix: raise the coverage-only timeout for the world allocation shards from 3600s → 5400s (50% headroom). Coverage-only; no effect on normal CI timing or runtime behavior.
Verified:
check-lint-cmake(gersemi) and codespell clean. The PR's ownCoverage (Debug)run exercises the new headroom.@codex review