Snapshot restore problem in ES 1.3.4


I encountered a problem with snapshot and restore, checking if somebody ran
into the same problem and if yes how they resolved it.

So this is what happened,

I did a snapshot of one our clusters and then restored that snapshot in
another cluster. When I did the restore there was a discrepancy of about 7
million documents from one index between the snapshot cluster and the
restored cluster(snapshot cluster had more documents in it). There were 3
other indexes that restored without any data loss. I did a shard
check(_cat/shards) and there was one primary shard and its 1 other replicas
that had 0 documents.

I deleted this index, and restored the index, but this time I set the
replica count to zero. This time around there was no data loss in the
restored index. The shard that was showing with 0 documents was now showing
with 7 million documents. Now I set the replica to 1 and after shards were
allocated again, there was again a data loss of 7 million documents. After
doing shard check, same shard and its replica had 0 documents.

Also the cluster from which I took the snapshot has ES1.2.2 installed and
the cluster in which snapshot was restored has ES1.3.4 installed on it.


Prateek Singhal

You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
To view this discussion on the web visit
For more options, visit