What is the best way to get the job failure errstr without using wait?

[SteVwonder]

flux.job.wait returns a jobid, failure boolen, and errstr if there was a failure. If you aren't using the wait interface (like in the new FluxEventExecutor class), then what is the best way to get the errstr? I couldn't find it in the main eventlog, guest.exec eventlog, or the KVS.

CC @jameshcorbett

[grondo]

The errstr is not stored anywhere, but is created by the job manager as it responds to a wait request when the job has exited.

https://github.com/flux-framework/flux-core/blob/master/src/modules/job-manager/wait.c#L72

The string itself is informational and is not meant to be part of job provenance, since it can be recreated from the finish event status.

A similar message (probably not exactly the same) is returned by flux job status by processing the eventlog, e.g:

$ flux job status -v ƒczpiyJKR
flux-job: ƒczpiyJKR: exited with exit code 1

[SteVwonder]

A similar message (probably not exactly the same) is returned by flux job status by processing the eventlog

Ah! Good to know. I thought the errstr contained additional information beyond the exit code. In that case, I think exit code is sufficient. Thanks @grondo!

[grondo]

Ah, well, technically the errstr will also describe job exceptions, so I suppose it isn't only the exit code, but all the information is indeed in the eventlog. I wonder if we should have a convenience function that produces the errstr as in flux job status or the job manager's decode_job_result()