Hi, I'm running SLURM version 14.11.9 and I'm trying to enable cgroup support for job accounting. I'm having trouble getting slurmd to properly track the job. The end result is that only the slurmstepd process is getting tracked. I'm getting errors in slurmd.log like:
[2015-09-30T15:03:33.582] [994] unable to open '/cgroup/memory/slurm/uid_7260/job_994/step_4294967294/task_0/memory.stat' for reading : No such file or directory [2015-09-30T15:03:33.582] [994] unable to get parameter 'memory.stat' for '/cgroup/memory/slurm/uid_7260/job_994/step_4294967294/task_0' [2015-09-30T15:03:33.582] [994] _prec_extra: failed to collect memory.stat pid 2556 ppid 2552 I'm not sure where the 'step_4294967294' is coming from. The only directory path under /cgroup is: /cgroup/memory/slurm/uid_7260/job_995/step_batch The job being launched is trivial: --- #!/bin/sh #SBATCH --ntasks=1 #SBATCH --time=00:15:00 sleep 600 --- It's launched with: > sbatch test.script At the moment, I have these in slurm.conf: --- ProctrackType=proctrack/cgroup TaskPlugin=task/cgroup JobAcctGatherType=jobacct_gather/cgroup --- and cgroup.conf contains: --- CgroupAutomount=yes CgroupReleaseAgentDir="/export/software/slurm/etc/cgroup" ConstrainCores=yes AllowedRAMSpace=100 ConstrainRAMSpace=yes AllowedSwapSpace=20 ConstrainSwapSpace=yes --- Any thoughts? Thanks, Kevin -- Kevin Hildebrand University of Maryland, College Park
