龙讯旷腾 pwmat Wiki

本页面只读。您可以查看源文件，但不能更改它。如果您觉得这是系统错误，请联系管理员。

~~NOTOC~~
===== q101: slurm fails to start inside the qflowcpu container =====

<WRAP lo>
==== Symptom ====
</WRAP>

Check the log file (run this inside the container):

<code bash>
 Check /var/log/slurm/d.log

 It contains the error "error: unable to mount freezer cgroup namespace: Operation not permitted"
</code>

In other words, if a submitted job just sits there doing nothing, or it looks like the slurm service never started, check this log file first using the method above and see whether this error appears.

<WRAP em>
==== Fix ====
</WRAP>

> Edit /etc/slurm/slurm.conf
> Change
> ProctrackType=proctrack/cgroup
> to
> ProctrackType=proctrack/linuxproc
>
> After saving the change, restart the container to apply it (run this on the host machine, not inside the container):
> <code bash>
docker restart qflowcpu
</code>

Once it's restarted, log back into the container and try submitting a job again to see if it works normally.

~~DISCUSSION:off~~