Thank you. I sync /etc/slurm.conf from slurmctld node then then re-run slurmdbd as root, which seems to solve the problem. Before that, "sacctmgr list clusters" complained missing /etc/slurm.conf .
Best, Jianwen > On Jan 11, 2019, at 12:33, Chris Samuel <ch...@csamuel.org> wrote: > > On 10/1/19 6:15 pm, Jianwen Wei wrote: > >> I am using SLURM 17.11 on CentOS 7. I find that QoS settings, say GrpNodes >> for an account, in sacctmgr requires restarting slurmcltd to take effect. > > This sounds like a configuration issue, and usually it's because slurmdbd > cannot connect back to the control daemon on the IP address and port it has > registered with it. > > If you do "sacctmgr list clusters" and then from the system running slurmdbd > try and connect with nc or telnet to the listed ControlHost on the listed > ControlPort you may find it doesn't work. > > Sometimes this is firewall related and sometimes this is because slurmctld > tells slurmdbd about an IP address that isn't reachable rather than one that > is. > > Hope this helps! > > All the best, > Chris > -- > Chris Samuel : http://www.csamuel.org/ : Melbourne, VIC >
