pc2 / cc-slurm-sync Goto Github PK

When I use cc-slurm-sync to start_job in cc-backend, the server notice that [WARNING] REST API: job without any resources/hosts.
I checked my config.json ,like this:
{ "slurm" : { "squeue": "/usr/bin/squeue", "sacct": "/usr/bin/sacct", "state_save_location" : "/var/spool/SLURM/StateSaveLocation" }, "cc-backend" : { "host" : "http://<my_host>:39000", "apikey" : "eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJpYXQiOjE2Njc4OTg1NDIsInJvbGVzIjpbImFwaSJdLCJzdWIiOiJzbHVybSJ9.tvSyPYBFRslez_hl9nT6nrKBSAQ9YYnPqagBuIg1AWkJi5YSoFWS2LbxuCVWAjwxOQDrTXIMHUlF4MSMpl2_CQ" }, "accelerators" : { }, "node_regex" : "^(n2(lcn|cn|fpga|gpu)[\\d{2,4}\\,\\-\\[\\]]+)+$" }
Is there any things wrong?

I forked the cc-slurm-sync project and used it in my cluster as the collection of job information with ClusterCcokkpit. During the setup, I found that original script was using slurm command to get the job state and information. This way is clear and understandable, but may consume more resources and connect with slurm server many times.

So I rewrite the script, change all slrum command to python API: pyslurm. This change also inferred the structure of slurmJobData, so you will see the conflicts. By the way, I use "logging" package instead of the "print".

I hope my rewriting will help your script more elignant and useful. The script is in https://github.com/SJTU-HPC/cc-slurm-sync. If you want, I'm willing to merge changes in cc-slurm-sync