During an intentionally peer restart, are its primary buckets always reassigned to the other cluster peers?
Namely I would like to know what happends to the primary buckets when the cluster is restarted using the command "splunk rolling restart cluster-peers" or when I apply a configuration bundle to the peers using the command "splunk apply cluster-bundle".
In these cases of cluster restart, are the primary copies of each peer reassigned and how it's done?
Primary buckets held by a given cluster peer will be re-assigned when the peer is restarted or becomes unavailable, regardless of the method used for restart. The purpose of a rolling restart (which 'splunk apply cluster-bundle' calls) is to orchestrate the restart in a way that allows this bucket primacy re-assignment to be done gracefully.
In the future, we hope to implement a 'cluster service mode' that will allow you to perform operations on your cluster without some of the cost currently incurred by a rolling restart - although this has more to do with bucket copies/rebuilds to respect rep/search factors than primacy re-assignment (we always need one primary bucket).