Ticket 3192

Summary: systemd slurmctld.service should start after slurmdbd.service and munge.service
Product: Slurm Reporter: Ole.H.Nielsen <Ole.H.Nielsen>
Component: ContributionsAssignee: Moe Jette <jette>
Status: RESOLVED FIXED QA Contact:
Severity: 4 - Minor Issue    
Priority: ---    
Version: 16.05.5   
Hardware: Linux   
OS: Linux   
Site: -Other- Alineos Sites: ---
Atos/Eviden Sites: --- Confidential Site: ---
Coreweave sites: --- Cray Sites: ---
DS9 clusters: --- HPCnow Sites: ---
HPE Sites: --- IBM Sites: ---
NOAA SIte: --- OCF Sites: ---
Recursion Pharma Sites: --- SFW Sites: ---
SNIC sites: --- Linux Distro: ---
Machine Name: CLE Version:
Version Fixed: 17.02.0-pre3 Target Release: ---
DevPrio: --- Emory-Cloud Sites: ---
Attachments: slurmctld.service patch for adding requirements

Description Ole.H.Nielsen@fysik.dtu.dk 2016-10-19 03:53:37 MDT 
Created attachment 3604 [details]
slurmctld.service patch for adding requirements

The RPM slurm-16.05.5-1.el7.centos.x86_64 file /usr/lib/systemd/system/slurmctld.service makes slurmctld start After=network.target.
However, the Munge and slurmdbd (in case you run the database) services are also required:
After=network.target slurmdbd.service munge.service
Patch attached.

Local custumization may actually be done by:
# cp /usr/lib/systemd/system/slurmctld.service /etc/systemd/system/
and changing the above line.
Comment 1 Moe Jette 2016-10-19 09:33:02 MDT 
Thank you for your contribution.

I modified all of the service files to not start Slurm daemons until after Munge is running.

I did not modify the slurmctld service file to start after slurmdbd. At most sites, the slurmdbd is running on a different computer (providing support for more than one cluster).

I also added your name to the contributor list.

The commit is here:
https://github.com/SchedMD/slurm/commit/660ffa2d8172dd30c6b5f4b04ee22d85f33222c3