Flaky test classes: the statement filter tests no longer depend on the runner's speed (Refs #5459) by erikdarlingdata · Pull Request #5504 · erikdarlingdata/PerformanceMonitor

StatementFilterAlertTests pin a judge budget that cannot run out; the over-budget path has its own test (#5459)

erikdarlingdata

committed

144f076

The install-tree wait test's holder process outlives a slow PowerShell start (#5459)

erikdarlingdata

committed

7e1f4da

DuckDB trim test waits for a trim cycle that really ran, not for one second of lock contention to pass (Refs #5459)

erikdarlingdata

committed

e33e442

Merge remote-tracking branch 'origin/fix/5459-w5-flakes' into lane-5459-w5c

erikdarlingdata

committed

15035b7

The CPU accountant's tests burn thread CPU time, not wall time, so a loaded runner cannot undercount them (Refs #5459)

erikdarlingdata

committed

296afe1

QueryStoreIntervalWideGridLiveTests waits out closed backends instead of a fixed sleep before counting scans (Refs #5459)

erikdarlingdata

committed

31feabf

Merge remote-tracking branch 'origin/fix/5459-w5-flakes' into lane-5459-w5b

erikdarlingdata

committed

7e21ffd

Revert "QueryStoreIntervalWideGridLiveTests waits out closed backends instead of a fixed sleep before counting scans (Refs #5459)"

erikdarlingdata

committed

c65cead

The CPU accountant's idle-side bounds allow one stray scheduler tick; the statement filter's slow-judge and three-incident tests run on a stepped budget clock (Refs #5459)

erikdarlingdata

committed

7bdce4e

Merge remote-tracking branch 'origin/fix/5459-w5-flakes' into w5d-work

erikdarlingdata

committed

9d76318

The snapshot worst-case test pins its statement-scrub session, so a loaded runner cannot spend the wall-clock budget; a spent session has its own test (Refs #5459)

erikdarlingdata

committed

b9f7979

The encrypt-password tests wait 180 s on their children, as the deploy probes do; a pin keeps the limit (Refs #5459)

erikdarlingdata

committed

4694f83

Navigation Menu

Repository navigation

Flaky test classes: the statement filter tests no longer depend on the runner's speed (Refs #5459) - #5504

Flaky test classes: the statement filter tests no longer depend on the runner's speed (Refs #5459)#5504
erikdarlingdata merged 12 commits into
devfrom
fix/5459-w5-flakes

Commits on Oct 8, 2026

StatementFilterAlertTests pin a judge budget that cannot run out; the over-budget path has its own test (#5459)

The install-tree wait test's holder process outlives a slow PowerShell start (#5459)

DuckDB trim test waits for a trim cycle that really ran, not for one second of lock contention to pass (Refs #5459)

Merge remote-tracking branch 'origin/fix/5459-w5-flakes' into lane-5459-w5c

The CPU accountant's tests burn thread CPU time, not wall time, so a loaded runner cannot undercount them (Refs #5459)

QueryStoreIntervalWideGridLiveTests waits out closed backends instead of a fixed sleep before counting scans (Refs #5459)

Merge remote-tracking branch 'origin/fix/5459-w5-flakes' into lane-5459-w5b

Revert "QueryStoreIntervalWideGridLiveTests waits out closed backends instead of a fixed sleep before counting scans (Refs #5459)"

The CPU accountant's idle-side bounds allow one stray scheduler tick; the statement filter's slow-judge and three-incident tests run on a stepped budget clock (Refs #5459)

Merge remote-tracking branch 'origin/fix/5459-w5-flakes' into w5d-work

The snapshot worst-case test pins its statement-scrub session, so a loaded runner cannot spend the wall-clock budget; a spent session has its own test (Refs #5459)

The encrypt-password tests wait 180 s on their children, as the deploy probes do; a pin keeps the limit (Refs #5459)