Hi,

One of our users had a job fail with the following error:

slurm-outfile: srun: error: Unable to create job step: Requested node
configuration is not available

Grepping the job ID out of the slurmcltd log file gives:

[2015-08-31T14:00:27.654] _slurm_rpc_submit_batch_job JobId=12345 usec=685
[2015-08-31T14:00:42.812] backfill: Started JobId=12345 on node[009,021,041]
[2015-08-31T14:01:02.361] _pick_step_nodes: requested nodes node[009,017,041] 
not part of job 12345
[2015-08-31T14:37:44.625] _job_signal: enter job JobID=12345 State=0x1 NodeCnt=3
[2015-08-31T14:37:44.626] deallocate_nodes:  job JobID=12345 State=0x8004 
NodeCnt=3

Can anyone explain how the nodes the requested nodes can be different to
the ones the job was started with?

This occurs with Slurm 14.11.8.

Cheers,

Loris

-- 
This signature is currently under construction.

Reply via email to