Comment 5 for bug 2015103

Based on your last comment:

> If I switch the leader of Keystone it does pass this hook and starts updating the endpoints like normal. It is possible that the certificates can not be updated for the leader units and that that causes the connection issue?

It sounds a bit like the underlying keystone service (running via apache2) hasn't been restarted. If this happens again, then try manually restarting the apache2 process first (on the leader) to see if this works around the issue.

I'll set the keystone part of this to incomplete for the moment.