-
Notifications
You must be signed in to change notification settings - Fork 5.8k
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
[Core][Worker_Pool] Wait for prestarted-workers for the first job and disable run_on_all_workers flaky tests #31836
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Stamped.
// This is a workaround to finish driver registration after all initial workers are | ||
// registered to Raylet if and only if Raylet is started by a Python driver and the | ||
// job config is not set in `ray.init(...)`. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Under what circumstances would the raylet be started by a Python driver?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
i think only local mode (starting ray with ray.init...
)
… disable run_on_all_workers flaky tests (ray-project#31836) Why are these changes needed? ray-project#30883 we changed the behavior of first driver connect to a raylet: previously the driver connection will wait for the prestarted workers to connect before returns, and after ray-project#30883 it no longer does so. This has caused some test flakiness for test_failure_4.py -k test_task_crash_after_raylet_dead_throws_node_died_error and test_ray_shutdown.py -k test_driver_dead We restore the previous behavior and fixes the test flakiness. Also ray-project#30883 added flakiness for run_functions_on_all_workers. we disable those flaky tests Signed-off-by: Andrea Pisoni <[email protected]>
Why are these changes needed?
#30883 we changed the behavior of first driver connect to a raylet: previously the driver connection will wait for the prestarted workers to connect before returns, and after #30883 it no longer does so. This has caused some test flakiness for
test_failure_4.py -k test_task_crash_after_raylet_dead_throws_node_died_error
andtest_ray_shutdown.py -k test_driver_dead
We restore the previous behavior and fixes the test flakiness.
Also #30883 added flakiness for run_functions_on_all_workers. we disable those flaky tests
Related issue number
Checks
git commit -s
) in this PR.scripts/format.sh
to lint the changes in this PR.