Not sure why it's corrupted, but it's not able to be snapshotted in that state. You may be able to snapshot the non-corrupted shards by setting partial to true in your Curator config (or with the regular API call).
That's a much deeper question. Is it the replica or the primary that is corrupt? If you don't know, then that makes it difficult to know which to address.
I'd ask a new question here in the forums, specifically targeting the shard corruption issue (rather than the snapshot failing) and see if one of the Elasticsearch core developers has a good answer for you.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.