(Answer to this old question)
Yes, possible as a workaround. In general, if you delete the corrupted bucket (splunk must be stopped, and remove the bucket and restart it again), Cluster Master should detect Replication factor is not met, and create a copy to a cluster peer.
In v6.5, there is action in fix-up bucket status page and select "remove/resync, etc"
If it is pre-v6.5, "remove_from_peer" REST call may be useful without stopping Splunk.
( ref: http://docs.splunk.com/Documentation/Splunk/6.5.2/RESTREF/RESTclusterExamples )
If that is failing, potentially somehow source bucket itself is corrupted. I would suggest to run;
splunk fsck repair --one-bucket --bucket-path=<bucket_path>
Hope this helps for people visiting this answer.
