Red Hat Bugzilla – Bug 1288211
Native HA deployment cannot find leader
Last modified: 2016-01-04 15:02:13 EST
Description of problem:
Attempted to restart atomic-openshift-master-api.service and atomix-openshift-master-controllers.service in order to confirm changes to the master-config.yaml, once the restart finished
Version-Release number of selected component (if applicable):
OSE 3.1, brand new installation as of 2015/12/2
Seems to be 100%
Steps to Reproduce:
1. Edit master-config.yaml
2. Restart atomic-openshift-master-api.service and atomix-openshift-master-controllers.service
the masters seem to have no "leader" for the system. The logs for the atomic-openshift-master-controllers.service all indicate
"Attempting to acquire controller lease as master-##thisid##
Lease openshift.io/leases/controllers owned by master-##differentid## at ...."
The owner id did not match any of the ids with which the masters are trying to use to become the leader. All three masters continue to attempt this until they either timeout or give up.
The logs continue with
"...unable to wait for lease expiration openshift.io/leases/controllers: 401: The event in requested index is outdated and cleared (the requested history has been cleared"
Everything to return to normal with the changes in the master-config.yaml made.
The change to the master-config.yaml was to the identity provider. Originally setup using htpasswd but changed to LDAP.