Comment 5 for bug 1796923

I'm not sure why but the nova-cloud-controller log entries stop on or before 2018-10-08 13:53:18:

$ tail -n1 nova-cloud-controller_*/var/log/juju/unit-nova-cloud-controller-*.log | grep -oE '^.*DEBUG'
2018-10-08 13:51:53 DEBUG
2018-10-08 13:52:03 DEBUG
2018-10-08 13:53:18 DEBUG

However, mysql does not seem to become clustered until 13:51:05

$ grep "All 3" mysql_2/var/log/juju/unit-mysql-2.log | head -1
2018-10-08 13:51:05 DEBUG juju-log cluster:31: All 3 percona units reporting clustered

and the hook that detects the cluster doesn't seem to complete until about 2018-10-08 13:54:14. This mean that the relation data is propagated to the nova-cloud-controllers until after 13:54:14.

On the surface it looks like the decision to abort the deploy was taken prematurely. Perhaps the logs were truncated before they were uploaded?

Please could you provide juju logs for nova-cloud-controllers for the period after 13:54? Alternatively could you try extending the timeout for the deploy ? (the nova-cloud-controllers only seem to have had 30mins between running their install hooks and the end of the logs)