0

I have a server that uses the shibboleth2 shibd daemon to allow remote authentication to a database.

The daemon reads a very large federation metadata file on starting/reloading (circa 34MB). This needs several minutes to process.

Previously it completed within the systemctl timeout limit of 90 seconds. It is noe exceeding this limit.

I have attempted to extend the systemctl timeout without success by editing /etc/systemd/system.conf in the following way:

DefaultTimeoutStartSec=600s
DefaultTimeoutStopSec=300s

This should override the default timeout of 90 seconds.

I then ran 

systemctl daemon-reload
systemctl daemon-reexec

to attempt to reload systemctl configuration itself to enable the new timeout.

However the timeout appears to have remained the same, and the command

systemctl start shibd

Times out after about 5 minutes

Are there any suggestions as to what is going on here? I have not retried simply rebooting the server yet as it is a production server and would affect multiple applications.

fred2
  • 97
  • 9
  • 1
    Are the values reported by `systemctl show shibd.service -p TimeoutStartSec` `systemctl show shibd.service -p TimeoutStopSec` `systemctl show shibd.service -p TimeoutSec` correct? – Mark Wagner Jul 25 '19 at 17:40
  • Hi @MarkWagner. Running those commands returned nothing at all, so I output all variable, and I got the following:` TimeoutStartUSec=5min` `TimeoutStopUSec=5min` `Result=timeout` `JobTimeoutUSec=infinity` `JobTimeoutAction=none` – fred2 Jul 26 '19 at 00:58
  • FYI - the current timeout appears to be 5 minutes, not 1.5 minutes as I said. However it is still too short for the metadata to be validated and import. – fred2 Jul 26 '19 at 01:11

1 Answers1

1

And excellent post explaining the process to extend the timeout is provided here:

https://unix.stackexchange.com/questions/227017/how-to-change-systemd-service-timeout-value/276785#276785

fred2
  • 97
  • 9