Thanks for your fast reply.
I tried 299 second and it works, than 301 seconds and it doesn't work.
I found out, that the max_switch_wait is one of the SchedulerParameters
(slurm.conf) and it is as you said, the default is 300 seconds.
Thank you for solving my problem.
Kind regards,
Danny
Am 26.01.2016 um 21:00 schrieb Antony Cleave:
Re: [slurm-dev] lua bug wait4switch
Try setting it less than 5 min.
If that works then it might be the max_switch_wait setting for the
cluster. I forget exactly where to set it now but I'm having a strong
sense of deja vu reading this and I remember thinking that's a very
short default when I changed it to 24 hours.
Antony
On 26 Jan 2016 10:08, "Danny Rotscher" <[email protected]
<mailto:[email protected]>> wrote:
Hello,
there seems to be a bug with the lua job submit plugin.
I want to change the variable wait4switch to a specific value, but
my changes are ignored by Slurm.
The value is always 5 minutes.
job_submit.lua.all:
[...]
function slurm_job_submit(job_desc, part_list, submit_uid)
[...]
job_desc.wait4switch=6000
[...]
scontrol output of the job:
JobId=391 JobName=sleep
UserId=rotscher(19426) GroupId=rotscher(19426)
Priority=25291 Nice=0 Account=rotscher QOS=normal
JobState=RUNNING Reason=None Dependency=(null)
Requeue=1 Restarts=0 BatchFlag=0 Reboot=0 ExitCode=0:0
RunTime=00:00:33 TimeLimit=08:00:00 TimeMin=N/A
SubmitTime=2016-01-26T10:38:39 EligibleTime=2016-01-26T10:38:39
StartTime=2016-01-26T10:38:39 EndTime=2016-01-26T18:38:39
PreemptTime=None SuspendTime=None SecsPreSuspend=0
Partition=sandy2 AllocNode:Sid=sandbox-login:13646
ReqNodeList=(null) ExcNodeList=(null)
NodeList=sandbox-node1
BatchHost=sandbox-node1
NumNodes=1 NumCPUs=4 CPUs/Task=1 ReqB:S:C:T=0:0:*:*
TRES=cpu=4,mem=40,node=1
Socks/Node=* NtasksPerN:B:S:C=0:0:*:1 CoreSpec=*
MinCPUsNode=1 MinMemoryCPU=10M MinTmpDiskNode=0
Features=(null) Gres=(null) Reservation=(null)
Shared=0 Contiguous=0 Licenses=(null) Network=(null)
Command=sleep
WorkDir=/tmp
Switches=1@00:05:00
Power= SICP=0
I created some debug output which prints out the
job_desc.wait4switch variable at the end of the job submit script.
The output shows me the correct value.
Thank you for any help!
Kind regards,
Danny
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Danny Rotscher
HPC-Support
Technische Universität Dresden
Zentrum für Informationsdienste und Hochleistungsrechnen (ZIH)
01062 Dresden
Tel.: +49 351 463-35853
Fax : +49 351 463-37773
E-Mail: [email protected]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
