Database model corruption due to previous compute node rename
Issue
In a previous version of OSP, compute hostnames were allowed to be “short” as in “not FQDNs”. A requirement that they become fully-qualified was introduced in OSP13, but also enforced in later z-stream releases of OSP10. Some deployments may have attempted to “rename” compute nodes from host
to host.domain
in place so as to satisfy the new requirement.
If this was done with any instances on the node, then the accounting for certain resources would have become corrupted and may be causing issues, even months or years later, and even after another major upgrade.
If this definitely happened on a cluster, then it is critical to perform the steps in the solution for affected instances and compute nodes. If it is not certain whether or not such a procedure was performed on the cluster, some steps below will help you make the determination, although they are not guaranteed to detect all cases where this may have happened.
Environment
- Red Hat OpenStack Platform 8
- Red Hat OpenStack Platform 9
- Red Hat OpenStack Platform 10
- Red Hat OpenStack Platform 13
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.