Leonardo back in production and scheduled operation on login nodes on December 6th

Dear Users,

Leonardo’s maintenance successfully concluded and the cluster is now back in full production. As anticipated, we changed the job environment to place the job’s tmpdir in a private /tmp mount point, with size of 10GB and accessible with the $TMPDIR variable. The area is mounted at the start of the job and unmounted at its completion. This change guarantees the nodes’ stability and prevents job’s unintended out-of-memory crashes.

We also inform that a network rewiring of the login nodes is scheduled on December 6th. The operation will cause a few seconds of disconnection, and will be announced via message on each login node prior to the operation. No impact on production is expected.

Best regards,

HPC User Support @ CINECA