Sync slurm job runner features with galaxy job runner implementation

[mvdbeek]

Nate Coraor
Yesterday at 10:43 PM
It's the same size on the JS2 side. The job hit the walltime:
slurmstepd: error: *** JOB 3561528 ON js2-large10 CANCELLED AT 2025-08-19T17:09:34 DUE TO TIME LIMIT ***

Nate Coraor
Yesterday at 10:44 PM
This happened with another job last week as well, Pulsar has never checked the DRM status of a job and I guess does not detect walltime terminations.

Which is about as bad as it gets, since the job appears to have succeeded from pulsar / galaxy's point of view.

[natefoo]

Unlike Galaxy, with Pulsar jobs the DRM state is completely ignored, only the tool exit_code file matters. Some relevant discussion here.

[natefoo]

Cleaning up some tabs and this is the linked discussion but tl;dr this should probably just be flipped to failed:

pulsar/pulsar/managers/base/base_drmaa.py

Line 49 in 3742d22

JobState.FAILED: status.COMPLETE, # Should be a FAILED state here as well