Skip to content
Snippets Groups Projects

Use status AW in myMdC if correction failed for some detectors in a run

Merged Thomas Kluyver requested to merge fix/mymdc-status-AW into master
2 unresolved threads

Description

We want to provide a clearer status when there are multiple detectors in a run and correction has failed for some but not all of them. The status 'AW' means 'available with warning'.

While trying to test this, I also noticed a problem with my recent changes to how we detect jobs are finished: we were checking the abbreviated Slurm state (e.g. CD) against the names of full states (COMPLETED), so I've tried to make this consistent. I can split this out if needed

How Has This Been Tested?

Deployed on max-exfl017, tried with myMdC test instance with p900113, r9981.

Relevant Documents (optional)

image

image

Types of changes

  • Bug fix (arguably)

Checklist:

  • My code follows the code style of this project.

Reviewers

@schmidtp @ahmedk @roscar

Merge request reports

Loading
Loading

Activity

Filter activity
  • Approvals
  • Assignees & reviewers
  • Comments (from bots)
  • Comments (from users)
  • Commits & branches
  • Edits
  • Labels
  • Lock status
  • Mentions
  • Merge request status
  • Tracking
55 59 :return: a dictionary indexed by slurm jobid and containing a tuple
56 60 of (status, run time) as values.
57 61 """
58 cmd = ["squeue", "--states=all"]
62 cmd = ["squeue", "--states=all", "--format=%i %T %M"]
  • 179 183 if str(r['job_id']) in statii:
    180 184 # statii contains jobs which are still going (from squeue)
    181 185 slstatus, runtime = statii[str(r['job_id'])]
    182 execn_ongoing_jobs.append(f"{slstatus}-{runtime}")
    183
    184 186 else:
    185 187 # These jobs have finished (successfully or otherwise)
    186 188 _, runtime, slstatus = slurm_job_status(r['job_id'])
    187 189
    188 190 finished = slstatus in STATES_FINISHED
    191 if not finished:
    192 short_state = STATE_ABBREVS.get(slstatus, slstatus)
    • This translates the most common states we'll see back to their short forms: PD & R. We have a .startswith('PD-') below to skip updates when all jobs are pending, so it's not entirely cosmetic.

    • Please register or sign in to reply
  • Thanks Karim!

  • merged

  • Thomas Kluyver mentioned in commit 3321be93

    mentioned in commit 3321be93

  • Philipp Schmidt changed milestone to %3.11.0

    changed milestone to %3.11.0

  • Please register or sign in to reply
    Loading