Comment 1 for bug 1852379

These happen when the underlying keystone service has restarted but the manager.py hasn't noticed/been restarted itself. I think there probably needs to be some logic in the manager.py to read the PID of the local keystone service when the manager.py first connects, and then if this error occurs during a an attempt to use the keystone API and the PID has changed, to re-connect to the new service, using the existing credentials. Thoughts?