Looking at logs from bug 1491228, it would appear that the first time that pacemaker goes to talk to the corosync daemon it gets denied. Per upstream docs [0], if the enable-acl property isn't explicitly enabled, then any user in the haclient group should have access. Since the hacluster charm doesn't explicitly enable the acl, I'd expect pacemaker to be running under the haclient/hacluster uid/gid.
The charms don't enable the acl and the package creates the hacluster user in the haclient group, so I suspect there's something additional going on here. For completeness, I'll attach the syslog in here. The pacemaker node did start after following JuanJo's workaround in comment #15. [0] - http://clusterlabs.org/doc/acls.html -- You received this bug notification because you are a member of Ubuntu Server Team, which is subscribed to lxc in Ubuntu. https://bugs.launchpad.net/bugs/1439649 Title: Pacemaker unable to communicate with corosync on restart under lxc To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/lxc/+bug/1439649/+subscriptions -- Ubuntu-server-bugs mailing list Ubuntu-server-bugs@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-server-bugs