Hi i was wondering if there was a way to route replicas to specific nodes in a cluster in the same way you can do shards with the allocation filtering. ??
My scenario is : I have a server with two blades in - each blade runs a couple of nodes - all of it clustered together.
I want to configure all my indexes to allocate Only on nodes on bladeA ...and all replications only to allocate on nodes on bladeB - so that I can lose a blade and still have a full set of data available.
If each blade was running just one node , ES would automatically allocate it as such - but it doesnt really help me if bladeA:node1 replicates to bladeA:node2 (Well, not for redundency)
Regards