If a split-brain has occurred between the cluster pods, we should prevent a scale down to 0 replicas from proceeding until the split has recovered.

To determine if a split-brain is present we should check the following:

• Are any of the Caches in a DEGRADED state?
• Does the CacheTopology/View have less members than spec.replicas?

We should also prevent the GracefulShutdown being initiated if a cache is in any state other than HEALTHY as a shutdown during HEALTHY_REBALANCE may also lead to data loss.