Skip to content

End of the annual maintenance#

We are pleased to announce the successful completion of our annual maintenance. As of today, February 19, 2024, all SCITAS services have been fully restored and are now operational.

Here is a brief overview of the upgrades and improvements implemented:

  • Central storage and backup Infrastructures: We have fully renewed our storage and backup infrastructures, doubling our capacity and introducing new services, including low-cost long-term archiving and file export options (NFS, SMB, S3). Please note that the archiving and backup service will be fully operational starting March 4, 2024. Consequently, there will be a two-week period without backup availability. However, it will still be possible to restore files created before February 12, 2024. A new pay-per-use billing system will also be implemented starting April 1, 2024.
  • Operating systems and software stack update: All clusters have been migrated to the RHEL 9.0 operating system, with updates to Slurm and other key software components to boost security and functionality. The software stack was recompiled for the new OS with minor adjustments. You’ll find the same modules as before. If you work with codes you installed, please make sure to recompile them before submitting any jobs.
  • Network infrastructure overhaul: Our interconnection network, which links the clusters both among themselves and with the external world, as well as the network dedicated to storage, has been upgraded to new infrastructures. This transition moves us from equipment provided by DSI to solutions directly managed by SCITAS. These changes guarantee improved throughput between the clusters and the central storage, as well as redundancy for all connections.

Access, data, and expected issues: - All clusters, including Helvetios, Izar, and Jed, are now back online and fully operational. Access to /home, /work, and /scratch data has been restored. - After updating the Slurm database, the scheduler removed jobs that were queued in a “pending” state.

Moving Forward: - We encourage you to verify the integrity of your data stored in /home and /work. - For information on how to utilize the new data capabilities, please refer to the updated documentation. - If you wish to access data stored on the clusters from your laptop via Samba or NFS, please reach out to us via email at 1234@epfl.ch, using “HPC: Scitas Data Export” as the subject line.

We’ve completed numerous tasks and implemented significant changes in a short time frame. Given the scope of these major updates, the possibility of encountering bugs exists. If you face any issues or need assistance, please contact us at 1234@epfl.ch with “HPC:” in the subject line. We appreciate your patience and cooperation during this period.

Happy computing!