Skip to content

SLURM job wait is None #358

Closed
Closed

Description

@j-woz

On Bebop, I think I am seeing cases where the output files indicate the job is done, but a short job.wait() is returning None, indicating that a final state has not yet been reached. Could this mean that job.wait() is caching an old scheduler status?

Activity

  1. hategan commented on Mar 10, 2023

    @hategan
    Collaborator

    Let me guess: the None status is in a different process than the one where the job was launched.

    It's because the status hasn't been updated yet.

  2. hategan commented on Mar 10, 2023

    @hategan
    Collaborator

    My bad. Sorry.

    Yes, it is possible. The queue is not polled continuously, so it's entirely possible to have a delay between when the job finishes and when the status is updated.

    The queue_polling_interval property in the executor config allows you to adjust the frequency with which the job status is updated.

  3. j-woz commented on Mar 10, 2023

    @j-woz
    ContributorAuthor

    Using queue_polling_interval works, PSI/J only returns None when not given enough time to poll.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions