Question

I'm setting up an Elasticsearch cluster using VMs in AWS.

I know that each node automatically tries to join an existing cluster which have the same cluster name in the same network.

However, I could not understand what the "same network" is.

Trying to understand about the same network, I found the concept of multicast and unicast. It seems to be the same concept between "same network" and "in the multicast group".

If so, how can I figured out what ip addresses are in the multicast group? and how can I add or remove some IP addresses in the multicast group?

I'm not so familiar with the concepts of network.

What I really want to know is that how can I figured out whether my nodes are in the same network or not? and how can I configure that.

Thanks in advance.

Was it helpful?

Solution

I am not a networking guy, but I believe they mean as long as they can see each other on a private network. Each node in a cluster wants to communicate internally with the other nodes on port 9300 by default.

So if you have one node on 10.20.10.5 (as an example) and one on 10.20.10.6, they should be able to "see" each other.

Also you can explicitly set the nodes in the cluster in the config file for Elastic.

For example in my elasticsearch.yml, I have something like this under the ### DISCOVERY ### section:

discovery.zen.ping.unicast.hosts: ["elastic-p1.domain.net:9300", "elastic-p2.domain.net:9300"]
Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top