Festus maintenance

On Sunday, June 14, between 4:00 AM and 10:00 AM, the Festus cluster will be unavailable due to major maintenance work. Your jobs will not be deleted and will automatically (re)start after the maintenance window. Some file systems may be inaccessible during this period. Please back up any data stored in /scratch that you still need.
When the upgrade is completed, students without a chairs account must use the “edu” partition and may access other resources only upon individual request within a supervised institutional project.

Use this command to check which slurm accounts are available for you:

sacctmgr show assoc user=$USER cluster=festus format=user,account%40


changes/fixes:

  • “default”-account as fallback doesn’t apply to student accounts anymore
  • “default”-account is limited to 4096 cores of simultaneously use
  • students have to use the “edu” partition
    • students may use other partitions only upon individual request within a supervised project
  • increase Priorityweight on financial share


details on what is upgraded:

  • beegfs 8.2 -> 8.3
  • rockylinux 10.1 -> 10.2
  • slurm 25.11.4 -> 25.11.6
  • Lmod 9.1 -> 9.2.2
  • cuda 13.1 -> cuda 13.2
  • pmix 5.0.8 -> pmix 5.0.10
  • add pmix-6 (6.1) support for slurm
  • rocm 7.2 -> 7.2.3
  • Firmwareupgrade on all servers/devices
  • OS updates Storageservers