Hello,
I am having a issue with QOSGrpCpuLimit error for an account.
We have an account (account name: chem_r) for a research group with 5 (eg. A,
B, C, D, E) group members.
We have 3 QOS partition
Name Priority GrpCPUs
---------- ---------- --------
normal 0
silver 0 4
gold 0 16
platinum 0 128
Account is assigned to Normal (default) parition
sudo sacctmgr modify account name=chem_r set qos=normal
Users are assigned to gold parition
sudo sacctmgr modify user A, B, C, D, E set qos=gold
If all five users submit the jobs with 4 cpus, the four are running and fifth
will receive error (QOSGrpCpuLimit).
Account have default partition and each user can run the jobs with 16 CPUs
independently.
Why slurm is adding CPUs from each user of same group.
How can I fix this issue?
Nilesh
