Unable to start instances with encrypted volumes after cluster restart

[Magnus Lööf]

I have enabled support for encrypted volumes in OpenStack Rocky. I use Ceph as storage backend.

There was no problem creating an encrypted volume and attaching it to an instance and filling it with data.

Later, there was a need for powering down the entire cluster for renovation work in the server room. The Ceph cluster was taken down according to these instructions:

https://ceph.com/planet/how-to-do-a-ceph-cluster-maintenance-shutdown/ (https://ceph.com/planet/how-to-do-a-c...)

Instances without encrypted volumes start nicely after the power up procedure.

However, instances with encrypted volumes do not start and this can be seen in the log files:

https://pastebin.com/mvMbJQRb

[Magnus Lööf (magnus-loof)]

I have determined that it is sufficient to reproduce by shutting down all kolla containers on the compute host to create the error.