You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
CI infrastructure improvements: heartbeat, dump checks, timeouts, logging (#15950)
* Change Heartbeat default interval from 5s to 60s
The 5-second interval generates excessive log output during CI test runs.
A 60-second default reduces noise while still providing periodic status
updates for diagnosing runner hangs and disk space issues.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Pass 60s heartbeat interval explicitly in CI workflow
Linux test steps were relying on the default interval. Now that the
default changed to 60s this is technically a no-op, but passing it
explicitly makes the intent clear and matches the Windows steps which
already specified 60s.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Move dump file check from final results to individual test jobs
The dump file check was in specialized-test-runner.yml and checked
artifacts/all-logs after downloading from sub-jobs. Move it to
run-tests.yml where it checks testresults/ directly in the same job,
matching the actual --crashdump/--hangdump output directory.
This catches crashes/timeouts earlier and in the correct job context.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Include timeout values in specialized test runsheets
Add testSessionTimeout and testHangTimeout MSBuild properties to the
JSON emitted by SpecializedTestRunsheetBuilderBase.targets, allowing
per-project timeout overrides to flow through the runsheet to the
CI workflow.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Clarify PR filtering log message for specialized test workflows
The old message 'filtering to single test project' was ambiguous.
Make it clear this is a sanity-check behavior for pull_request events.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Clamp heartbeat interval to >= 1s to prevent tight loops
Addresses review feedback: reject zero or negative interval values
that would cause a tight loop or crash in Task.Delay.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Narrow dump file check to hang dumps only
Crash dumps (e.g. *_crash.dmp) can be produced during process cleanup
even when all tests passed — this is benign. Only fail on hang dump
files (*hangdump*) which indicate the test runner timed out and tests
may not have completed.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
0 commit comments