So I've two 0.90 clusters, in two datacenters, and I've an rsync script on a single node in the new cluster that scrapes the _cluster/state of the old cluster, and rsyncs shards from the appropriate nodes.
Because of the migration window and ES version, I don't have the luxury of snapshots. My plan was to run the rsync script repeatedly upto the migration date to keep the deltas small. Then put the old cluster into read only mode, and do a final sync. Once the sync is done, start ES on the sync node and join it to the new cluster.
My issue is that I can only get some synced indices to come up within the cluster and I'm not sure why only some.
Is there some setting to trick ES 0.90 to bring up unknown indexes shards found on disk into the cluster?