openshift/cluster-etcd-operator

Optimization: cluster member controller 2

Closed this issue · 0 comments

CMC2 maintains the invariant that at any given time in bootstrap process only 1 member will be unhealthy. Because of this invariant, only 1 etcd pod will be in running but not ready state. To take advantage of this we should modify the code here to select pod that is running but not ready. This will make sure that only the pod that is waiting will get added as a member.