Maintenance of the WORK file system on SuperMUC-NG

Hoechstleistungsrechner

The second high performance file system WORK of SuperMUC-NG (phase 1 and 2) needs to undergo maintenance from

Saturday 7. of February 8:00 until Wednesday 11. of February 18:00
--> prolonged to Wednesday 18, 10:00

Update 16.02 18:00 We are currently testing the stability of WORK with IO test and plan to resume regular operation on Wednesday morning.

13.02. 14:00: We are still encountering network instabilities between filesystem and nodes. The issue is under investigation. As a safeguard we extended the maintenance period until Monday. Please continue to submit jobs that depend only on SCRATCH as described below.

-——

During the maintenance no jobs will be started by SLURM that require the WORK partition, which is assumed by default. Jobs that do not require WORK can still run, but need to be marked accordingly with

#SBATCH --licenses=work:0

in your SLURM job script. For further documentation see

https://doku.lrz.de/list-of-slurm-constraints-and-its-usage-10747322.html

-——

12.02 10:30 The security update to SLURM has been applied. Running jobs seemingly were not affected, but please check in case. Maintenance of the WORK filesystem continues with performance tests. Maintenance has been prolonged until Friday.