Cluster Controller at p55cc1

-SLURM updated from version 15.08 to 16.5
-Kernel updated to version 3.10.0-514.10.2.el7.x86_64

Both clusters are in pre-maintenance state due to slurm upgrade.

Maxtime is set to 1 hour.

Shutdown will be initiated without any further notification.

Both clusters are ready to use now.

Due to new kernel both clusters will be rebooted as soon as all queues have been drained.

The partition max. runtime has been set to 1 hour to allow monitoring jobs execute.

Depending on the currently running jobs' max. time settings, the reboot will happen at latest one week from now.

The reboot will initiated without any further notice!

Both clusters have now 'fake' compute node #99 which is dedicated to Grid jobs (including cluster monitoring).

This means that all 'normal' compute nodes (titan nodes 1-9 and pleione nodes 1-32) can be freely used for real work without disturbing monitoring jobs.

I.e. you can take over all the nodes from normal, small, big, all (whatever, except grid) partitions.

Nodes pl17, pl22 and pl99 IB cables fixed.

Cluster is ready to use.

The cluster is in pre-reboot state and max runtime has been decreased to 1 hour. Once all partitions has been drained the cluster will be rebooted.

Titan reboot scheduled for 23.1 already done.

Partitions maxtime restored to one week so the cluster is ready to use.

Pleione reboot scheduled for 23.1 already done.

Partitions maxtime restored to one week so the cluster is ready to use.

Due to new kernel both clusters need reboot.

All partitions' max runtime is set to 1 hour until the reboot. So, don't panic if (when) your (new) job is put on hold until the reboot and max runtime restored.