Which node(s) should be used for indexing vs Queries

(Seth Klock) #1

Hello everyone. I was wondering if someone could give me a little guidance on which servers in a cluster should be used for indexing and querying. I have a total of 7 nodes. One is the primary master another is the secondary master and a data node. The rest are just data nodes. We're using the NEST client. For queries we round-robin the requests to the 2 master nodes. For Indexing we use the primary master node. Does this sound right? I know we could setup "client" nodes to just manage the query requests. I'm considering that, but even if we do that I'm not sure which node(s) should be used for indexing.

(Luca Cavanna) #2

When you say master nodes, do you mean nodes that are master eligible only? Or are those nodes that also hold data?
In general I would round-robin both index requests and search requests among data nodes.

(Christian Dahlqvist) #3

Based on your description it looks like you have 2 nodes that are master eligible. In order to allow a majority to be formed even when one of the master eligible nodes is unavailable, you should aim to always have exactly 3 master eligible nodes.

(system) #4