Hi Jonathon, In non-root, it always returns "slurmd is stopped" when it actually failed to check /var/run/slurmd.pid which is permission denied. Me too, I guess that reason for not using a simple "status slurmd" because for supporting different platforms.
Cheers. On Thu, Feb 11, 2016 at 5:29 PM, Jonathon A Anderson < [email protected]> wrote: > I suspect the slurm init script doesn’t use the `status` command because > it is specific to upstart, and not necessarily available in all init > systems (e.g., in sysvinit). > > The slurmd pid file on one of our compute nodes appears to be > non-root-readable.[1] What error were you having when trying to check the > slurmd status before modifying the init script? > > ~jonathon > > > [1]: # stat /var/run/slurmd.pid > File: `/var/run/slurmd.pid' > Size: 5 Blocks: 8 IO Block: 4096 regular file > Device: 1h/1d Inode: 108568 Links: 1 > Access: (0644/-rw-r--r--) Uid: ( 0/ root) Gid: ( 0/ root) > Access: 2016-02-10 12:34:39.959156252 -0700 > Modify: 2016-02-08 11:23:10.517122210 -0700 > Change: 2016-02-08 11:23:10.517122210 -0700 > > > > On Feb 10, 2016, at 10:56 PM, jupiter <[email protected]> wrote: > > > > Hi Jonathon, > > > > I figured out, the problem is not the root owner ship but the way > /etc/init.d/slurm implements the "service slurm status", it checks the pid > file and caused permission issue. Why did it simply run "status slurmd" > which works perfectly? > > > > I've modified the status and works fine now, thanks for your response. > > > > status) > > prog="${0##*/}d" > > status ${prog} > > ;; > > > > > > > > > > On Thu, Feb 11, 2016 at 1:05 PM, Jonathon A Anderson < > [email protected]> wrote: > > slurmd must run as root because it forks and execs processes on behalf > of other users using the job owner’s uid. > > > > I don’t understand what trouble you’re having monitoring slurm with > nagios. Could you give an example of what you’re trying to do, what you > expect it to do, and what it does in stead? > > > > ~jonathon > > > > > > > On Feb 10, 2016, at 6:30 PM, jupiter <[email protected]> wrote: > > > > > > Hi, > > > > > > I am running slurm on CentOS 6. One thing I've just noticed is that > the slurmctld is running under the user slurm but the slurmd is running > under the root. Not quite sure why those daemons in different owner ships, > any inside explanation please? Further looked at both slurm daemon > configure in /etc/init.d and slurm.conf, they are identiccal, how could > they behave differently? Anyway, can the owner ship of slurmd be changed to > the user slurm? > > > > > > The problem I've got now is I am running nagios monitoring via ssh, it > can check all other application daemon status, but it always failed to > check slrum daemon status due to the slurmd root access restriction. > > > > > > Thanks. > > > > > > > > > > > >
