The 20th of December at around 4PM, we upgraded Slurm to version 23.11.1. The reason why we did that outside of a maintenance period of the cluster is because there was a critical security discovered in Slurm that had to be fixed ASAP. From user side, this should be transparent and we hope you even didn’t noticed the upgrade. We are investigating if the issue some of you had with GPUs is related.
Thanks for your understanding