Hi,
I am using qlaunch to indirectly launch fireworks. I need to set the job to the ‘med’ partition, but strangely this is not working.
When running a slurm job manually the priority is set by prepending the slurm command #SBATCH -p med argument at the head of the script to be run. For example,
#!/usr/bin/env python
#SBATCH -p med
#SBATCH --job-name=XX
#SBATCH --output=job.out
#SBATCH --error=job.err
#SBATCH --time=16:00:00
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=16 -c 2
To do this with the qlauncher I have tried to change the qlauncher configuration file to include the partition key.
_fw_name: CommonAdapter
_fw_q_type: SLURM
ntasks: 32
partition: med
rocket_launch: rlaunch -w /home/dda/run_fw/sn_opt/my_fworker.yaml -l /home/dda/run_fw/sn_opt/my_launchpad.yaml singleshot
cpus_per_task: 1
ntasks_per_node: 1
walltime: '05:00:00'
queue: null
account: null
job_name: null
logdir: /home/dda/run_fw/sn_opt/logs
pre_rocket: source activate fw37
post_rocket: null
I have tried adding quotes around the ‘med’, yet in both situations the same (incorrect) FW_submit.script that is generated does not have the partition argument in it. The generated FW_submit.script looks like this
#!/bin/bash -l
#SBATCH --ntasks=32
#SBATCH --ntasks-per-node=1
#SBATCH --cpus-per-task=1
#SBATCH --time=05:00:00
#SBATCH --job-name=FW_job
#SBATCH --output=FW_job-%j.out
#SBATCH --error=FW_job-%j.error
source activate fw37
cd /home/dda/run_fw/sn_opt
rlaunch -w /home/dda/run_fw/sn_opt/my_fworker.yaml -l /home/dda/run_fw/sn_opt/my_launchpad.yaml singleshot
# CommonAdapter (SLURM) completed writing Template
I have double checked and found that the line #SBATCH --partition=$${queue} is present in the SLURM_template.txt file, yet for some reason it is not being parsed correctly.
Does anyone have any idea what could be going wrong?
Thank you