Since every customer is unique, I would create a support case for this issue, provide diags and the list of buckets that is corrupted. Provide how the index cluster is set-up. e.g. single site with 3 indexer and replication factor of 2.
Depending on the replication factor, if the bucket is corrupted and it's a replicated bucket, rb_epochendtime_epochstarttime_bucketid_guidofindexer
You might be able to move the bucket to the temp directory. Enable maintenance mode on cluster master, restart of the indexer, then disable maintenance mode and the cluster master should tell the index cluster to create another copy.