'schedselect' missing in 'qstat -xf'

Users/Site Administrators
1

Hello,

We submitted a quick test job with the following #PBS directives:

#PBS -q def-short
#PBS -N simpletest
#PBS -j oe
#PBS -l select=1:mem=10gb:ncpus=4:mpiprocs=4
#PBS -l walltime=04:00:00

However, we received the following comment on the queued job:

comment = Can Never Run: Insufficient amount of resource: qlist (compute,gpuReady != compute,gpuHost,gpuReady,lmHost)

Furthermore, while reviewing qstat -xf, we notice the schedselect is missing:

…
Rerunable = True
Resource_List.mem = 10gb
Resource_List.mpiprocs = 4
Resource_List.ncpus = 4
Resource_List.ngpus = 0
Resource_List.nodect = 1
Resource_List.place = free
Resource_List.select = 1:mem=10gb:ncpus=4:mpiprocs=4
Resource_List.walltime = 04:00:00
…

Based on the default_chunk in the queue configuration, we expect the ‘qlist’ attribute added to the schedselect:

create queue def-short
set queue def-short queue_type = Execution
set queue def-short Priority = 140
set queue def-short acl_host_enable = False
set queue def-short resources_max.walltime = 24:00:00
set queue def-short default_chunk.qlist = compute
set queue def-short default_chunk.qlist += gpuReady
set queue def-short resources_available.ngpus = 0
set queue def-short max_run_res.gpuHost = [o:PBS_ALL=2]
set queue def-short enabled = True
set queue def-short started = True

What configuration issue might result in this?

Thanks,
Siji

2
  1. did you try to run the qstat -xf as the root user ?
  2. I hope qlist is appended to the resources: line in the sched_config file and pbs_sched is kill -HUP
  3. increase the sched_log level and check the scheduler logs for any reason.

Thank you