We are attempting to test out rack property in our future architecture
layout. To verify rack configuration though, I stood up two ES 0.20.4 VM’s
and they are configured as Master/Data nodes each.
Node 1:
node.rack_id: r1
cluster.routing.allocation.awareness.force.rack_id.values: r1,r2
cluster.routing.allocation.awareness.attributes: rack_id
Node 2:
node.rack_id: r2
cluster.routing.allocation.awareness.force.rack_id.values: r1,r2
cluster.routing.allocation.awareness.attributes: rack_id
Everything worked as expected. The primary shards where allocated on one
node while the replicas where on the other node. No issues.
Our configuration that we are investigating is spliting out ES into
separate servers as follows:
Master Eligiable Only ( MEO-1 )
Master Eligiable Only ( MEO-2 )
Client Only ( CO-1 )
Client Only ( CO-2 )
Data Only ( DO-1 )
Data Only ( DO-2 )
Data Only ( DO-3 )
Data Only ( DO-4 )
I picked DO-1 & DO-3 to be r1 and DO-2 & DO-4 to be r2. Configured the
MEO-1&2 to have all the properties defined above. No matter how I configure
any nodes, ES continues to rebalance the shards across all the DO’s. It
does not put the primary shards in DO-1/3 and replicas in DO-2/4 or vise
versa.
Any thoughts?
--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.