Hi,
It seems corosync does not signal successful start in any way.
This causes the following problem on boot for me:
Starting Corosync Cluster Engine...
[ OK ] Started Corosync Cluster Engine.
Starting Corosync Dbus and snmp notifier...
[ OK ] Started Corosync Dbus and snmp notifier.
Starting dlm control daemon...
[ OK ] Started dlm control daemon.
Starting cluster LVM daemon...
[FAILED] Failed to start cluster LVM daemon.
See 'systemctl status clvmd.service' for details.
And in the logs:
Nov 02 11:45:13 dev1n1 clvmd[2751]: Can't initialise cluster interface
clvmd can be manually started later without problems.
The problem would be probably solved if 'Started Corosync Cluster
Engine.' would be delayed until the cluster is initialized.
When running corosync in the background (forking) this could be
signalized by the parent process returning only after the cluster is
ready.
When running in the foreground (that is how I use it under systemd) the
sd_notify() function from the sd-daemon library could be used.
I can provide a patch for the latter, but I would need some help.
Which point of the Corosync startup should be considered the 'cluster
ready' state? Should I consult the clvmd sources first?
Greets,
Jacek
_______________________________________________
discuss mailing list
discuss@xxxxxxxxxxxx
http://lists.corosync.org/mailman/listinfo/discuss
