SLURM Interactive Desktop job not launching desktop, output.log is empty

[kumarcraja (C Raja)]

Dear team,

Configuring OOD first time. Unable to launch interactive Desktop with slurm batch job. Also output.log is empty.

Please kindly do the needful to diagnose and resolve the same.

Following is the configuration file contents.

cat /etc/ood/config/clusters.d/iiser.conf

v2:
metadata:
title: “iiser”
login:
host: “compute01.hpcdomain.com”
default: true
job:
adapter: “slurm”
cluster: “iiser”
bin: “/usr/bin/”
conf: “/etc/slurm/slurm.conf”
batch_connect:
copy_environment: true
basic:
script_wrapper: |
module purge
%s
set_host: “host=$(hostname -A | awk ‘{print $1}’)”
vnc:
script_wrapper: |
export PATH=“/opt/TurboVNC/bin:$PATH”
export WEBSOCKIFY_CMD=“/usr/local/bin/websockify”
set_host: “host=$(hostname -A | awk ‘{print $1}’)”
acls:

• adapter: “group”
  groups:
  • “user1”
  • “user2”
  • “root”
  • “ood”
    type: “whitelist”

cat /etc/ood/config/apps/bc_desktop/iiser.yml

title: “iiser XFCE Desktop”
cluster: “iiser”
attributes:
desktop: “xfce”
batch_connect:
template: “vnc”
websockify_cmd: “/usr/local/bin/websockify”

Images showing the web interface stats

[jeff.ohrstrom]

I’d check the output from sacct to see what the exit code is or what Slurm is doing. It seems like it’s being scheduled and immediately failing - maybe out of memory issue or some odd shell thing.

In any case, because sacct shows historical jobs you can see the exit code of the job and that exit code/and or slurm error should point you in the right direction.

[kumarcraja (C Raja)]

Hi jeff,

sacct command output here

slurm job exit code is normal 0.
Able to submit normal batch mode jobs and running fine.
Interactive Desktop job is completing with in few seconds, but it is not showing any output and output.log is empty.

Hearty Thanks,
C Raja

[jeff.ohrstrom]

OK… That doesn’t make much sense that it exits 0 immediately and without anything in the logs. I see it’s a fairly new cluster… there’s something here we’re missing that’s obvious and now I think I see it.

You need the %s in the script_wrapper. I see you have it for basic - but not for vnc. This is the important bit. %s is literally the script that we provide. It is the script that you’re wrapping. Since you didn’t add it - we didn’t!

Look at the file job_script_content.sh and you’ll see there’s basically nothing there (this is the file that’s submitted to Slurm).

[kumarcraja (C Raja)]

Dear Jeff,

Now issue is resolved. Able to launch remote desktop.

Added %s in vnc template section as below

vnc:
  script_wrapper: |
    export PATH="/opt/TurboVNC/bin:$PATH"
    export WEBSOCKIFY_CMD="/usr/local/bin/websockify"
    %s
  set_host: "host=$(hostname -A | awk '{print $1}')"

Hearty Thanks, going the extra mile for all of us who depend on you. I simply couldn’t be more grateful…

C Raja