Fil des news

Something is broken or hang in our networking. We are working on it.

EDIT 11:19 - I still don't now precisely what happened, but every login node, visu node and compute node need a reboot.

EDIT 11:40 - @login OK, @visu OK

reboot of compute nodes ongoing, every running jobs will be lost.

EDIT 14:30 - @compute is restarting

EDIT 15h00 - @compute is fully restarted

We may have to deal with extrem mesures in the next few weeks/months, including :

• half, or all, of computing powered off,
• complete power outages, including storage, with J-1 warnings,
• 9 hours a day of job suspension.

This is due to electricity prices and possible power outages.

Situation is still very fluid, decisions have not been made yet.

“Hope for the best, but prepare for the worst.”

A storage node for /scratch/Cral is having issues. This result in a partial unavailability (not loss) of data.

Some updates in our slurm documentation, see:

• https://meso-centres-lyon.pages.in2p3.fr/psmn-rtd/clusters_usage/submitting.html#slurm-commands (slurm basics)
• https://meso-centres-lyon.pages.in2p3.fr/psmn-rtd/clusters_usage/computing_resources.html#features (features/constraints)
• https://meso-centres-lyon.pages.in2p3.fr/psmn-rtd/clusters_usage/partitions_overview.html (features/constraints)

A Power Distribution Unit (and a fan) unexpectedly died yesterday evening. Cascade group s92node[67-78] was brutally shutdown, running jobs have been lost.