You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
parallelcluster 3.9.1 and 3.11.1
I see many files in /tmp since the start of the head_node of the form:
-rw-r----- 1 slurm pcluster-slurm-share 1088 Nov 18 09:14 tmp.VBWqTAz4SS
-rw-r----- 1 slurm pcluster-slurm-share 262 Nov 18 09:05 tmp.zMJAymfADb
-rw-r----- 1 slurm pcluster-slurm-share 276 Nov 18 09:01 tmp.4FNDMLk4rC
-rw-r----- 1 slurm pcluster-slurm-share 282 Nov 18 08:59 tmp.xVx3sST9n3
-rw-r----- 1 slurm pcluster-slurm-share 282 Nov 18 08:55 tmp.a23WPhksqt
-rw-r----- 1 slurm pcluster-slurm-share 473 Nov 18 08:41 tmp.D3MvXoHL1g
-rw-r----- 1 slurm pcluster-slurm-share 1691 Nov 18 08:40 tmp.zYS9m1CU3j
-rw-r----- 1 slurm pcluster-slurm-share 1488 Nov 18 08:40 tmp.mCOXXmqOHt
-rw-r----- 1 slurm pcluster-slurm-share 1488 Nov 18 08:39 tmp.di3dHuc923
-rw-r----- 1 slurm pcluster-slurm-share 265 Nov 18 08:37 tmp.AXTQoAX1hL
-rw-r----- 1 slurm pcluster-slurm-share 265 Nov 18 08:37 tmp.45sPSvBlrB
I would expect slurm (slurmctld since we're on the head node?) to clean up and not leave crumbs.
Contents of the file seem to relate starting jobs and are of the form:
parallelcluster 3.9.1 and 3.11.1
I see many files in /tmp since the start of the head_node of the form:
-rw-r----- 1 slurm pcluster-slurm-share 1088 Nov 18 09:14 tmp.VBWqTAz4SS
-rw-r----- 1 slurm pcluster-slurm-share 262 Nov 18 09:05 tmp.zMJAymfADb
-rw-r----- 1 slurm pcluster-slurm-share 276 Nov 18 09:01 tmp.4FNDMLk4rC
-rw-r----- 1 slurm pcluster-slurm-share 282 Nov 18 08:59 tmp.xVx3sST9n3
-rw-r----- 1 slurm pcluster-slurm-share 282 Nov 18 08:55 tmp.a23WPhksqt
-rw-r----- 1 slurm pcluster-slurm-share 473 Nov 18 08:41 tmp.D3MvXoHL1g
-rw-r----- 1 slurm pcluster-slurm-share 1691 Nov 18 08:40 tmp.zYS9m1CU3j
-rw-r----- 1 slurm pcluster-slurm-share 1488 Nov 18 08:40 tmp.mCOXXmqOHt
-rw-r----- 1 slurm pcluster-slurm-share 1488 Nov 18 08:39 tmp.di3dHuc923
-rw-r----- 1 slurm pcluster-slurm-share 265 Nov 18 08:37 tmp.AXTQoAX1hL
-rw-r----- 1 slurm pcluster-slurm-share 265 Nov 18 08:37 tmp.45sPSvBlrB
I would expect slurm (slurmctld since we're on the head node?) to clean up and not leave crumbs.
Contents of the file seem to relate starting jobs and are of the form:
{"jobs":[{"extra":null,"job_id":17836,"features":null,"nodes_alloc":"sp-r7a-m-dy-sp-8-gb-1-cores-40","nodes_resume":"sp-r7a-m-dy-sp-8-gb-1-cores-40","oversubscribe":"NO","partition":"sp-8-gb","reservation":null},{"extra":null,"job_id":17837,"features":null,"nodes_alloc":"sp-r7a-m-dy-sp-8-gb-1-cores-41","nodes_resume":"sp-r7a-m-dy-sp-8-gb-1-cores-41","oversubscribe":"NO","partition":"sp-8-gb","reservation":null}],"all_nodes_resume":"sp-r7a-m-dy-sp-8-gb-1-cores-[40-41]"}
Is this a bug, feature, or known issue? Should I be cleaning up head_node /tmp/ files older than N days?
The text was updated successfully, but these errors were encountered: