To Reproduce Steps to reproduce the behavior:
glance-api containers (both external and internal) stay in CrashLoopBackOff loop. From glance-api logs it looks like glance is unable to start workers and bind to proper port (glance.common.wsgi doesn't log message about starting workers). The only two messages logged consistently are

2025-04-27 17:45:54.963 2 INFO glance.async_ [-] Threadpool model set to 'EventletThreadPoolModel'
2025-04-27 17:45:55.919 2 WARNING keystonemiddleware.auth_token [-] AuthToken middleware is set with keystone_authtoken.service_token_roles_required set to False. This is backwards compatible but deprecated behaviour. Please set this to True.

Not sure if it is related, Ceph RBD backend is used

must-gather is attached to the case

Expected behavior
When startup problems are faced, glance should log meaningful errors

Bug impact
New RHOSO deployment is blocked, customer is unable to run tests and add workloads

Known workaround
None