Service | Incident status | Start Date | End Date |
---|---|---|---|
Cedar | Closed |
Planned Outage - Arrêt planifié
Cedar will be unavailable from 2025-03-31 to 2025-04-01 . During this
time, you will not be able to log in or run jobs on the cluster. Any
jobs that are running at the time of the outage will be terminated and
you will need to re-submit them once the cluster is back online.
We have extended the outage to April 2 at 10am. We're working to ensure services are tested and available after the work was completed. At 10am we intend to begin restoring access to the cluster and slowly bringing compute resources back online.
======
Cedar ne sera pas disponible du 2025-03-31 au 2025-04-01. Durant cette période, les utilisateurs ne pourront pas se connecter ni exécuter des tâches sur Cedar. Toute tâche en cours au moment de l’arrêt de service sera arrêtée et devra être soumise à nouveau une fois les travaux terminés.
Update:
The outage has been completed, we're still bringing more compute nodes online to process jobs, but the cluster is operational.
Notes:
Slurm Upgrade
- Upgraded from 24.11.0 to 24.11.3
Home has been migrated to new storage
- This should be transparent to users
- New Home Performance
- Designed to match or exceed previous Home's performance.
Nearline has been migrated to new storage as well
- First time you request a file from nearline, you must use:
lfs hsm_restore {filepath}
- This only affects the first time that a file is recalled
- Normal operations of nearline will continue for any new files
- If you do a normal cp and notice your file is 0 bytes, it means you need to
do lfs hsm_restore {filepath}
Updated by Adam Spencer on