-
Notifications
You must be signed in to change notification settings - Fork 44
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Retest only test that fail and not entire package #368
Comments
The option The use case of the option is to check for flaky tests. So passing |
Woops, I typo-ed, I meant to reference |
For CMake packages the option For Python packages using colcon-core/colcon_core/task/python/test/pytest.py Lines 126 to 137 in 2a12b22
|
@ruffsl https://app.circleci.com/pipelines/github/ros-planning/navigation2/3553/workflows/67da27fb-e87b-46d9-b33b-840882c9bb19/jobs/13927/steps this is one such job. You can see it fails the first time on nav2_system_tests after about 10 minutes (about as long as it takes to run all of the tests), then it re-runs and take another 10 minutes to fail. No single test takes 10 minutes so that was re-running the full suite. Feel free to ignore the actual failure from this PR CI run. This is just the first demonstration I could find. |
The log shows a completely separate second invocation of This kind of repetition has nothing to do with the re-running of tests based on the arguments |
This may have gotten lost in the line nose thanks to Looks this is a side effect from ros-navigation/navigation2#1284 . I'll close this and move discussion there. |
Correct me if I'm mistaken, but it seems that
colcon test --retest-until-fail N
colcon test --retest-until-pass
will rerun all tests for a failing package, as opposed to only the specific tests in the package that failed. Could this be changed so the results from a previous job be used as a block list to scope which specific tests need to be repeated, not merely the package. This could help speed up CI pipelines with flaky tests.@SteveMacenski, could you link to a previous job that exhibits the behavior you described to me offline.
The text was updated successfully, but these errors were encountered: