Nodes gpu001 and gpu003 down on bamboo

[Brian.Pulfer]

Primary informations

Username: pulfer
Cluster: bamboo

Description

Nodes gpu001 and gpu003 are down / drained.

Steps to Reproduce

salloc -n 1 -c 1 --time 01:00:00 --nodelist gpu001
or
salloc -n 1 -c 1 --time 01:00:00 --nodelist gpu003

Expected Result

New shell session in gpu001

Actual Result

The allocation just hangs.

salloc: Requested partition configuration not available now
salloc: Pending job allocation 293409
salloc: job 293409 queued and waiting for resources

[Adrien.Albert]

Hi Brian,

gpu001 was drained for:

[Slurm] Create public-interactive-gpu partition

Dear Users, Cluster: Bamboo, Yggdrasil We would like to inform you about a recent change to the GPU partition on the system. The gpu001 partition has been modified from debug-gpu to public-interactive-gpu. This change introduces the following constraints for all users: Key Changes: Partition Name: gpu001 has been moved to the public-interactive-gpu partition. Constraints: 1 Job per User: Each user is limited to running a single job at a time. 1 GPU per Job: Each job will be allocated only o…

gpu003 is drained to prepare the next week maintenance.