Just upgrade the slurmdbd, then shutdown the cluster, install the new  
slurm and restart the daemons. No jobs should be lost.

Quoting Mario Kadastik <[email protected]>:

>> Hi, upon startup slurmctld changes its working directory to where  
>> the log file is.
>> If the log file is:
>> SlurmctldLogFile=/var/tmp/slurm/slurmctld.log
>> the working directory is /var/tmp/slurm. Assuming your slurmctld  
>> core dump for whatever reason the core file should be there.
>> The directory should be writable by the SlurmUser since slurmctld  
>> changes it real user id to that user.
>
> bah, seems I had forgotten to set the log file so as I said the log  
> was in /var/log/messages. I've reconfigured slurm to now write it  
> separately, but it does mean that if it did core dump, then not in  
> /var/log.  As it's configured now, then a new crash would leave the  
> core.
>
> Also, is the upgrade to latest 2.5.x a live upgrade? Or do I need to  
> drain the cluster first?
>
> Thanks,
>>
>
> Mario Kadastik, PhD
> Researcher
>
> ---
>   "Physics is like sex, sure it may have practical reasons, but  
> that's not why we do it"
>      -- Richard P. Feynman
>
>

Reply via email to