I'm posting two separate messages on related subjects just to make each one focused and simpler to answer. ( Another one is about s3 gateway backup)
My questions are for the people who uses local gateway backup in EC2.
If I have a large cluster, lets say 20 nodes with multiple large indexes, 10-15 shards in each index, and 2 replicas, how do I decide which nodes I need to backup? I don't need to backup each and every node with 3 replicas and so many shards, do I ? It looks like a lot of wasted storage to me.
How do you deal with transient nature of nodes and their IP addresses in EC2 in restore? I guess, it's simple during backup, just backup a local node. But are there any issues with changing IPs with restore?
I understand, my backup will be a bunch of folders, one per original node. How do I restore a cluster from that? Do I create a new node per a backup folder and restore each of them manually?