[slurm-users] Fwd: Problem installing slurm on computer cluster

[Philip Kovacs]

>Forgive me for saying this. I do have a bit of experience in building HPC systems.>Distro supplied software packages have improved a lot over the years.>But they do tend to be out of date compared to the latest versions of (say) Slurm.
It is actually a great deal of work to package Slurm for some distros.   It is a big package with many moving parts.  Very difficult to test comprehensively on multiple architectures and with hundreds/thousands of possible configurations.   That said, I think I do a fairly good job with the Fedora releases.  It is volunteer work for most of us.
Phil   

    On Thursday, November 16, 2017 9:08 AM, John Hearns <hearnsj at gmail.com> wrote:
 

 Forgive me for saying this. I do have a bit of experience in building HPC systems.Distro supplied software packages have improved a lot over the years.But they do tend to be out of date compared to the latest versions of (say) Slurm.I really would say you should consider downloading and installing from the vendors site.The same thing goes for compilers, MPI, and many software packages.
Yes, the distro supplied debs or RPMs are easy to install and will be tested against that distro.But (again) they will eb installed locally on the nodes. So again with compilers, MPI , Python modules....you tend to instlal these on a network shared drive so that you have one central install.




On 16 November 2017 at 14:47, E V <eliventer at gmail.com> wrote:

You may need to install a systemd override file if you have some of
the system config left over, it has the path set to /usr/bin/. Example
for slurmd, slurmctld and slurmdbd are the same just changing the
names:

cat /etc/systemd/system/slurmd. service.d/override.conf
[Service]
ExecStart=
ExecStart=/usr/local/sbin/ slurmd $SLURMD_OPTIONS

On Wed, Nov 15, 2017 at 12:37 PM, Bruno Santos <bacmsantos at gmail.com> wrote:
> Hi everyone,
>
> I am currently trying to install slurm to serve as a job scheduler for a
> research institute. I have installed Debian stretch and initially installed
> and configured slurm from the repos.
> HoweverI then tried to play with a different server serving as node and
> realized that due to the different versions of debian the controller and the
> daemon where running different versions of slurm and so not working.
> I have since done apt-get remove --purge and tried to install slurm from
> src. But it seems that the old configuration is still stuck somewhere as
> when I try to run:
>
> #systemctl enable slurmctld
> Synchronizing state of slurmctld.service with SysV service script with
> /lib/systemd/systemd-sysv- install.
> Executing: /lib/systemd/systemd-sysv- install enable slurmctld
> Failed to enable unit: Unit file /etc/systemd/system/slurmctld. service is
> masked.
> # whereis slurmctld
> slurmctld: /usr/local/sbin/slurmctld
> # slurmctld
> bash: /usr/sbin/slurmctld: No such file or directory
>
> Any idea what could be going wrong?
>
> Thank you very much in advance.
> Best,
> Bruno
>
>





   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.schedmd.com/pipermail/slurm-users/attachments/20171116/5ede8a03/attachment.html>