RWTH High Performance Computing (HPC)

As part of the accounting data fixes, we are regenerating the missing accounting data since 06.01.2026

Aufgrund von Slurm-Problemen ist jupyterhub aktuell nicht erreichbar. An einer Lösung wird gearbeitet.

We are encountering an issue where Slurm does not send emails at the end of jobs. We are working on workarounds.

Issues with IPV6 and sendmail were fixed. Emails should be sending correctly.

Coordinator roles in Slurm are currently missing and we are working on restoring them.

We are aware if issues starting Slurm jobs on certain accounts and sacct. We are working on a solution.

A configuration error was made in the job accounting of the HPC cluster on December 12th, which resulted in incorrect billing values being generated. This issue has been resolved as of December 30th, and all jobs submitted from this point onward will be billed correctly. We are currently working on correcting the erroneous billing for affected jobs.

Values have been corrected. Changes should be visible in short

During the maintenance we will upgrade the Slurm scheduler to the newest version.
We will also work on resolving recent issues with sacct and Slurm Accounts not being available at Job submission.
Finally, we will setup a tuned and improved job priority configuration for the submission queues, with the goal to improve QoS based on user
feedback we have received.

To achieve this, all jobs still pending in the queue by the time the maintenance starts will be removed by us. These will have to be resubmitted by users. We ask users to please take measures to track and resubmit the jobs if there is a need for it.

The Upgrade of Slurm and required accounting fixes is taking longer than expected and will extend the Maintenance period by one extra day.