Deployment times out due to nova-conductor never starting

[Ben Nemec]

I've been seeing this on a semi-regular basis lately. The deployment times out, and when you start looking at logs it appears nova-conductor never started, which causes nova-compute to time out. I haven't been able to figure out why conductor isn't starting though. There are no conductor logs, and the only failures I see appear to be related to https://bugs.launchpad.net/tripleo/+bug/1604708 However, I see those failures in successful jobs too, so not sure it's related.

Example log: http://logs.openstack.org/54/344454/2/check-tripleo/gate-tripleo-ci-centos-7-ovb-ha/ceb83dc/logs/overcloud-novacompute-0/var/log/nova/nova-compute.txt.gz

[Ben Nemec (bnemec)]

I'm going to bump this to critical because it's causing a lot of our CI runs to fail spuriously, and take a long time in doing so. Basically it's wasting a ton of CI resources right now.

[Ben Nemec (bnemec)]

I haven't noticed this happening in a while, so I'm going to mark it fixed. Can always reopen if it pops back up.