Migration from PBS/Torque to SLURM

Migration from PBS/Torque to SLURM

The computing cluster of the Department of Mathematics ‘Tullio Levi-Civita’ is migrating from the control software ‘PBS/Torque’ to SLURM workload manager. The basic idea is keep the two cluster active and migrate the computing host from the old cluster to the new one.

Migration plan:

start: the slurm commands are available on the hosts labsrv7 (new) and labsrv8. From these machines you will be able to submit jobs to the handful of computing hosts (dellcuda1, gpu01, hpblade07, hpblade0201) presently aggregated to the new SLURM cluster.

In the next days the hosts still aggregated to the old PBS/Torque cluster will be removed from the available queue (waiting for the termination of the current jobs) , upgraded and aggregated to the new SLURM cluster.

At the end of the process only labsrv0 (submitting host) and one computing host (still to be determined) will remain in the PBS/Torque cluster for verification tasks.

The time of the migration depends from the termination of the corrent jobs but we plan to migrate on or two computing host for week