Has anyone had an upgrade to their ES cluster inadvertently delete all indices? We have several clusters, and a rolling update worked on all of them, except 1 cluster. The nodes were all 1.5.2, and on that cluster we did one thing differently; instead of terminating a node (one at a time) and bringing up the new node (running the newer version, 1.6.0) we had brought on an additional node. So in the 3 node cluster, we had 3 nodes running 1.5.2, and terminated 1 of them, and brought up 2 nodes at the new version 1.6.0. That [shockingly] caused all indices to be deleted. Any ideas?
Do the indexes exist in /your_es_data_dir/nodes/#/indices/index_name? Elasticsearch won't delete any indexes unless you tell it to but possibly something is wrong with your indexes so they aren't visible. I'd check the Elasticsearch logs for any errors.