Stabilize warm-refresh timing test under host load #66

New issue

Closed

opened 2026-07-24 09:51:23 +00:00 by coilyco-ops · 1 comment

coilyco-ops commented

2026-07-24 09:51:23 +00:00

Member

Observation

TestWarmRefreshIsFastAndReuses failed during the full ward --audit-override-dirty exec test gate, then passed on an unchanged rerun.

Evidence

Failed run: warm refresh took 327.8094ms, budget 250ms.
Unchanged rerun: internal/launch passed in 2.163s and the full gate passed.
The failure appeared while all Go packages were running and several package test times were elevated.

Candidate causes

Fixed wall-clock threshold is sensitive to Windows host scheduling and concurrent package load.
Cold filesystem or process startup variance may be included in the measured warm path.
The assertion may need a deterministic reuse signal separate from a generous performance regression bound.

Done when

The test still verifies warm reuse without failing under ordinary loaded-host variance.
The test does not hide a real cache or refresh regression.

## Observation `TestWarmRefreshIsFastAndReuses` failed during the full `ward --audit-override-dirty exec test` gate, then passed on an unchanged rerun. ## Evidence * Failed run: warm refresh took `327.8094ms`, budget `250ms`. * Unchanged rerun: `internal/launch` passed in `2.163s` and the full gate passed. * The failure appeared while all Go packages were running and several package test times were elevated. ## Candidate causes * Fixed wall-clock threshold is sensitive to Windows host scheduling and concurrent package load. * Cold filesystem or process startup variance may be included in the measured warm path. * The assertion may need a deterministic reuse signal separate from a generous performance regression bound. ## Done when * The test still verifies warm reuse without failing under ordinary loaded-host variance. * The test does not hide a real cache or refresh regression.

coilyco-ops commented

2026-07-24 10:03:07 +00:00

Author

Member

Second unchanged occurrence while validating #67 on 2026-07-24: TestWarmRefreshIsFastAndReuses measured 268.1778ms against the 250ms budget. All affected person and roster packages passed. The deterministic pre-commit, palette, lint, build, and serial smoke checks also passed.

Second unchanged occurrence while validating #67 on 2026-07-24: `TestWarmRefreshIsFastAndReuses` measured `268.1778ms` against the `250ms` budget. All affected person and roster packages passed. The deterministic pre-commit, palette, lint, build, and serial smoke checks also passed.

coilyco-ops referenced this issue

2026-07-24 10:03:39 +00:00

Add placeholder seats for AOSH-selected OSS harnesses #67

coilyco-ops referenced this issue from a commit

2026-07-24 11:40:45 +00:00

test: make warm refresh reuse deterministic (closes #66)

coilyco-ops closed this issue

2026-07-24 11:40:45 +00:00