Update to Rocky Linux 9 and decommissioning Skylakes

On Thursday, July 17, we will begin rolling upgrades to all partitions of Hydra and Anansi to Rocky Linux 9.5. Running and queued jobs will not be impacted. On Monday, July 21, we will start decommissioning the skylake partition and reduce the skylake_mpi partition to 16 nodes.

These operations are part of the ever ongoing strategy to keep VUB’s Tier-2 HPC infrastructure modern and up to current standards. As many of you know, the Hydra cluster has an heterogeneous hardware configuration that is regularly upgraded.

Goodbye skylake, Hello zen5

After years of dependable service, the skylake nodes will be stepping down. On Monday, July 21, we will fully decommission the skylake partition and reduce the skylake_mpi partition to just 16 nodes.

The recently added zen5 nodes are already in full production and provide a lot more compute capacity than the old skylakes and can pick up the slack with much more overall performance.

Rolling Upgrade to Rocky Linux 9.5

Starting Thursday, July 17, we will begin a rolling upgrade of all partitions to Rocky Linux 9.5. This is a major system upgrade from version 8.10, but no action should be needed on your part. The new zen5 nodes have already been running version 9.5 smoothly since their release, and the 2024a toolchains have been rebuilt for the new system version.

Older software should continue to work without issue but if anything breaks, just drop us a line at VUB-HPC Support and you can still rely on the 16 nodes in skylake_mpi which will be kept on the previous 8.10 version for a little while.

Upgrades are done partition by partition, so your running or queued jobs won’t be affected, though you may notice slightly longer queue times during the process.

Contact us at VUB-HPC Support if you have any comments or questions about these changes.