ERROR: t2800-jobs-cmd.t - exited with status 137 (terminated by signal 9?)

[chu11]

I've been seeing this a lot of late when running make -j16 check.

ERROR: t2800-jobs-cmd.t - exited with status 137 (terminated by signal 9?)

My assumption is after #2744, rc3-job is just taking a bit too long. This was removed in that PR:

test_expect_success 'cleanup job listing jobs ' '
        for jobid in `cat job_ids_pending.out`; do \
            flux job cancel $jobid; \
            flux job wait-event $jobid clean; \
        done &&
        for jobid in `cat job_ids_running.out`; do \
            flux job cancel $jobid; \
            flux job wait-event $jobid clean; \
        done
'

I think the default grace timeout is used, which would be 1 second, so I can totally see flux job cancelall taking a bit too long.

I'll add a little something to test_under_flux to see if that helps.

[grondo]

Any reason not to just use flux job cancelall here?

[chu11]

You mean re-add the above, but just use flux job cancelall? I was initially thinking of just upping the grace timeout, but adding a flux job cancelall would probably work.

Any preference? Each seems equally simple to add.

[chu11]

actually, pausing on a fix given #2733

[grondo]

Ah, sorry my comment above was out of context. I had forgotten that code was already removed.

[chu11]

With #2733 merged didn't see this issue after 10 runs, i'm going to assume it's fixed.