This issue has already been seen. Here’s what I find on the other report :
"datanodes had failed and they could not rebalance. This was resolved but subsequently rebalances failed on index nodes. pskill indexer on each of the index nodes allowing them to come back up ready between each one, allowed the rebalance to complete successfully.
What has worked to allow the rebalance to succeed is after the rebalance fails, restarting the affected service on each node. … had them do a rolling restart of the data nodes."
So you can try that, otherwise, if you have an enterprise license, contact customer support and work through the issue with them.