Comment 3 for bug 1808598

We have discussed this issue in the IRC channel today. Given that the change that caused this (https://review.openstack.org/#/c/611530/8) is covering an edge case it seems that this is something we should be able to work around.

In discussion we talked about changing the migration to catch the failure and check for the DB type. In the case that the DB is mariadb then make the failure non-fatal. There is some precedence for doing something like this: https://github.com/openstack/cinder/blob/master/cinder/db/sqlalchemy/migrate_repo/versions/085_cinder_init.py#L782