RWTH High Performance Computing (HPC)

Dear Users,
We had to set our Slurm queuing system down due to a potential security vulnerability.
New jobs will not run and the queue will not be terminated, new jobs might still be queued but is not guaranteed..
Currently running Jobs might still be able to finish to run, but we might have to kill some long running jobs.
The login nodes should still be available for normal data work.
All this might change without notice during the day to accommodate for further mitigation and resolution actions as needed.

The vulnerability has been fixed and the cluster is back in operation.

Due to the limitted availabiloty of the GPFS, the connection to the cluster could not be established.

The problem has been resolved and the cluster is available again.

We would like to inform you that a security issue (CVE) has been discovered in the Munge software, which allows for the potential exposure of the Munge key. This key is critical for user authentication in Slurm jobs. To address this security vulnerability, we have rolled out a new version of Munge.
Although we assess the likelihood of the key being compromised as very low, we will be conducting maintenance on the cluster to exchange the key.
We kindly ask all users to review their jobs. If you notice any unknown jobs in your list, please delete them and inform us immediately.
For users with extremely sensitive data in their directories, we offer to temporarily disable job submission for your account upon request. Your account will be re-enabled after the maintenance is completed.

On February 5, 2026, from 09:00 to 10:00 AM, electrical work will be carried out that affects the air-cooled rack of CLAIX2023 as well as the Infiniband switches of CLAIX2023.We would like to point out that the systems and switches are redundantly connected, so normally no service interruption is expected. However, in rare cases, brief interruptions may occur due to a misconfiguration.