Elasticsearch query result consensus across replica nodes

(Nathan Baulch) #1

I'm working on a project that uses Elasticsearch as a primary data store for sensitive data that needs to be resilient to corruption. The indexes are split into 5 shards with 2 replicas across 3 machines.

Is there a built in way to send queries to all 3 nodes and check for consensus? Is this something I need to code explicitly in my client or can the data node I happen to connect to handle this for me?

Also, does Elasticsearch perform any parity/integrity checks on source and index data, aside from the potentially slow startup check when index.shard.check_on_startup is enabled? Or does this need to be handled by a fault tolerant FS or in hardware with RAID-5/10?

(Mark Walkom) #2

The former.

We did add checksums to the entire stack (lucene and Elasticsearch) a while back.

(system) #3

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.