Blanca Slurm configuration updated

Update:

Due to an oversight, jobs that attempted to start after this configuration change when into effect were held with the status "launch failed requeued held." This was due to the fact that the default Slurm "spool" directory had not been created, whereas the previous configuration specified a non-default location. This left Slurm with nowhere to store job scripts on the target node when they were dispatched, causing the jobs to fail to start.

The default spool directory has now been created on all Blanca compute nodes, and the affected jobs have been released to run again.

-

The Blanca Slurm configuraiton was updated today to address problems with backfill scheduling in some very specific edge cases. For most users there should be no apparent change; but if you have any issues running jobs on Blanca, please notify rc-help@colorado.edu.

Thu, 12 Apr 2018 -0600