SLURM commands

[verdurin]

At the suggestion of @fgeorgatos - let me know exactly what would be needed to add SLURM support; this is from hazy memory...

List all nodes:

[root@slurm000 ~]# sinfo -N | head
NODELIST   NODES PARTITION STATE
ca000          1  compute* alloc
ca001          1  compute* alloc
ca002          1  compute* comp
ca003          1  compute* comp
ca004          1  compute* alloc
ca005          1  compute* alloc
ca006          1  compute* alloc
ca007          1  compute* mix
ca008          1  compute* alloc

List jobs running on a specific node:

[root@slurm000 ~]# squeue --nodelist ca002
  JOBID PARTITION                           NAME     USER ST            TIME  NODES   CPUS MIN_MEMORY NODELIST(REASON)
2555843   compute                        s-35798   user1 CG      2-16:21:03      1      1         4G ca002
2555841   compute                        s-19415   user1 CG      2-16:21:03      1      1         4G ca002
2547777   compute                           bash  user2  R      3-00:02:32      1      1         4G ca002
2527888   compute                           bash   user3  R      4-02:18:18      1      1         4G ca002
2546116   compute                           bash  user4  R      3-02:09:31      1      1         4G ca002
2612665   compute                           bash  user5  R         3:33:45      1     20        20G ca002
2618855   compute                           bash  user6  R           16:16      1      1         4G ca002
2611854   compute                           bash  user7  R         4:49:33      1      1         4G ca002
2544926   compute                           bash  user8  R      3-03:02:25      1      1         4G ca002

[fgeorgatos]

thank you @verdurin : this is a good basis to get SLURM info collection going;
i think the last command implies a bit of recursion in the output of the commands (first ask list of nodes, then iterate through them)

then, we'll need a list of jobs in the queue to x-check what is running and what is pending. to be continued