i have the Problem, that in the init script of slurm 15.08.0 both pid files (slurmd and slurmctld) are mentioned.
# processname: /usr/sbin/slurmd
# pidfile: /var/run/slurmd.pid
#
# processname: /usr/sbin/slurmctld
# pidfile: /var/run/slurmctld.pid
The script itself checks correctly with "scontrol show daemons" which services should be started, but systemd waits for all pid files/services.
This leads to an timeout error on compute nodes, because only the slurmd started.
At the Moment i use two different startup scripts for compute & login nodes, where the slurmctld lines are removed.
Is this problem allready known ? Is there a way to use the same init script for all nodes without these problems with systemd ?
Best
Sven
----
Dipl.-Chem. Sven Siebler
Hessisches Kompetenzzentrum für Hochleistungsrechnen
Universität Kassel
ITS Raum 250
Mönchebergstraße 11
D-34109 Kassel
Tel.: +49 561 804-7759
[email protected]
www.hpc-hessen.de
smime.p7s
Description: S/MIME Cryptographic Signature
