Dear all,
we have found a small bug in slurm version 14.3 and 14.11. Everytime a
new user is added we see the following behaviour: When displaying the
user associations you can see the correct default qos, but it is not
known when submitting jobs. The newly created user always has to specify
the default qos in the job script. However, when modifying the account
and resetting the QOS and default QOS, the user needs no longer to
specify the default qos. See a the slurm commands, that we used, below.
Thank you for your help!
best regards
Markus
====================================================================================
#Create two qos:
sacctmgr -i add qos name=qos_test_a grpnodes=10 maxwall="3-0"
description='test_a'
sacctmgr -i add qos name=qos_test_b grpnodes=10 maxwall="3-0"
description='test_b'
#add account with default qos:
sacctmgr add account test_def_qos QOS=qos_test_a,qos_test_b
DefaultQOS=qos_test_a
#add partition with these qos to slurm.conf:
PartitionName=test_qos Default=NO MinNodes=1 AllowGroups=ALL Priority=1
DisableRootJobs=NO RootOnly=NO Hidden=NO Shared=NO GraceTime=0
PreemptMode=OFF ReqResv=NO AllowAccounts=test_def_qos
AllowQos=test_qos_a,test_qos_b
Nodes=n01-[010-012,045,094-096],n02-[001,006,013],n03-[025,066,071]
#add user to previously created account:
sacctmgr -i add user 70032 DefaultAccount=test_def_qos
#user associations:
[markus@l31 ~]$ sacctmgr show user `id -u` withassoc
format=user,defaultaccount,account,qos%40s,defaultqos%20s
User Def Acct Account
QOS Def QOS
---------- ---------- ----------
---------------------------------------- --------------------
70032 test_def_+ test_def_+
qos_test_a,qos_test_b qos_test_a
#JOB test1:
------------------------------------------
#!/bin/bash
#SBATCH -J jobname
#SBATCH -N 1
#SBATCH --tasks-per-node=16
#SBATCH --partition test_qos
sleep 600
------------------------------------------
[markus@l31 slurm]$ sbatch job.sh
sbatch: error: Batch job submission failed: Job has invalid qos
#JOB test2
[markus@l31 slurm]$ cat job.sh
------------------------------------------
#!/bin/bash
#SBATCH -J jobname
#SBATCH -N 1
#SBATCH --tasks-per-node=16
#SBATCH --partition test_qos
#SBATCH --qos qos_test_a
sleep 600
------------------------------------------
[markus@l31 slurm]$ sbatch job.sh
Submitted batch job 23508
#REDEFINE qos's
sacctmgr modify account test_def_qos set QOS=qos_test_a,qos_test_b
DefaultQOS=qos_test_a
Modified account associations...
C = vsc3 A = test_def_qos of root
C = vsc3 A = test_def_qos U = 70032
Would you like to commit changes? (You have 30 seconds to decide)
(N/y): y
#JOB test3
[markus@l31 slurm]$ cat job.sh
------------------------------------------
#!/bin/bash
#SBATCH -J jobname
#SBATCH -N 1
#SBATCH --tasks-per-node=16
#SBATCH --partition test_qos
sleep 600
------------------------------------------
[markus@l31 slurm]$ sbatch job.sh
Submitted batch job 23511
--
=====================================================
Dr. Markus Stöhr
Zentraler Informatikdienst BOKU Wien / TU Wien
Wiedner Hauptstraße 8-10
1040 Wien
Tel. +43-1-58801-420754
Fax +43-1-58801-9420754
Email: [email protected]
=====================================================
