We are also facing the same issue. @anil or anybody care to explain and define the accurate behaviour? For sure to me the documentation is confusing and is not aligned to what we are observing.
Is there a hack/workaround to not rebuild the indexes?
@ingenthr @ddash I have been researching the same topic, having the same issue (map-reduce view index rebuilds after failover even though we have view index replicas). Like other Couchbase users in this thread, we have found no adequate explanation for this.
Could somebody familiar with the matter at Couchbase elaborate?