Reading ES documentation I couldn't find a confirmation, if a client node is also useful durning indexing process.
Client node role is to support search operations, so data node doesn't has to do that.
But if client node is also useful during indexing?
Then, i.e., logstash output is set only to client nodes, that spread data over data nodes.
If so, is it better than output directly to data nodes?
Another question, can I put client node and data node on the same machine (asumming, there is 128 GB RAM, 4 Cores Processor, etc.), is it good way of configuring claster?
Regards,
Wojtek