Access restored for Leonardo and DCGP back in production

Dear users,

The vulnerability issue that impacted production yesterday, April 30th, is resolved also on Leonardo. The cluster is now accessible but it is operational on the DCGP partition only. The subnet manager issues causing the node communication problems on Leonardo have also been solved and the system state is currently stable. The booster partition is expected to resume production operations by tomorrow.

Please note that, in order to apply the fix, it was necessary to interrupt all running jobs on the cluster to allow a full restart of all compute nodes. We therefore recommend that you check the status of your work now that access has been restored.

Best regards,
HPC User Support – CINECA