Hello,
I've question to the parameter --nnodes of sacct.
sacct -a --nnodes=Min-Max
I understand the line above, that sacct shows all jobs which have a node
count between min and max including the min and max values.
But as you could see in the following section, it doesn't work as I suggest.
Moe,
On 7/7/15 7:04 PM, Moe Jette wrote:
-- Backfill scheduler: The configured backfill_interval value
(default 30
seconds) is now interpretted as a maximum run time for the backfill
scheduler. Once reached, the scheduler will build a new job queue and
start over, even if not
Bill and Bruce,
We are just in order to get fairshare. I'm not gonna do that in production
though, that sounds dangerous.
You mean you don't want to be the guinea pig?!
The reservations in the database are only for historical purposes, they
don't get read in from the slurmctld. The DBD
The backfill scheduler will get to the end of the queue if it can do
so in 30 seconds (or whatever you have backfill_interval configured to
be). The sdiag command will report actual scheduler run times. The
cycle times are in units of microseconds.
Quoting Martins Innus
Danny,
I think if you add more to your output you will likely see that the results
that display with 1 node are actually just the batch parts of another job,
but they show up on their own line. I am not a slurm expert, so there could
be other reasons for the 1s showing up in your output, but this
Last night for 8 hours our MySQL server was offline due to a storage
failure. As far as I can tell from the slurmctld logs jobs continued to be
started and complete successfully. The only errors I saw were in the
slurmdbd logs, as expected. What, if any, impact on a SLURM cluster will
there be