Info

The next scheduled cluster maintenance is on January 21st, 2027.

To ensure the stability and security of our compute clusters, C&CZ performs regular maintenance.

Affected Clusters

The maintenance applies to the following clusters:

  • Science cluster for FNWI
  • Coma cluster for Astro (of FNWI)
  • MLP cluster for Letteren

Maintenance Schedule

Maintenance is scheduled every six months. While we would prefer to perform these updates more frequently, we have chosen this interval to minimize disruptions to ongoing scientific calculations and research work.

What Maintenance Entails

Our update strategy is designed to keep the systems secure while minimizing unplanned downtime:

  1. Daily Security Updates: All cluster nodes use unattended-upgrades to automatically install critical security updates on a daily basis.
  2. Pre-Maintenance Updates: Shortly before the scheduled maintenance window, we configure unattended-upgrades to also install regular system updates.
  3. Maintenance Day: On the designated maintenance day, we reboot the nodes. This is necessary to apply the latest kernel updates and ensure the system is running the most current, stable version of the operating system.

User Communication and Impact

We strive to provide ample notice so that users can plan their jobs accordingly.

Announcements

Maintenance windows are announced via:

  • The CPK page on this website.
  • The CPK mailing list.
  • A short notification in the MOTD (Message Of The Day) upon logging into the nodes.

Scheduling and Downtime

For nodes managed by SLURM, we create a reservation for the maintenance window. This prevents the scheduler from starting new jobs that would overlap with the maintenance, avoiding the risk of jobs being killed prematurely.

Users are encouraged to check the maintenance schedule before submitting long-running jobs.