Looking at backing up my elasticsearch index on a small three node cluster. The _snapshot api states a need for a shared mount. If that is not a viable option what's the best/efficient way to backup. The nodes run on VMs so I could snapshot. Can I potentially use the _snapshot api locally(on one node), but having to restore on each node manually? What are my options?
The snapshot API is designed to work with some kind of shared storage abstraction like a shared filesystem (NFS is fine) or HDFS or something like that.
If you use a hard disk snapshot as a backup you'll backup a lot of duplicate information and restoring from backup will be like recovering from a full cluster power outage. It can take some time but it ought to get the job done. The snapshot API is much more flexible in what it keeps and for how long.
Can the snapshot API work without a shared file system? How would the backup and restore work under that scenario?
No. Every node must be able to write to that dir.
Unless you have only one node but I'm sure you don't want to stop your other nodes 
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.