Elasticsearch Cluster Status Yellow

(Vilas Reddy Podduturi) #1


I have created a cluster with two nodes.
Config on server1:

cluster.name: app.elasticsearch.prod
node.name: app.elasticsearch.prod.prod01

Config on server2:

cluster.name: app.elasticsearch.prod
node.name: app.elasticsearch.prod.prod02

I have stopped elasticsearch one server at a time couple of times for config modifications.
In head plugin of each server, I see the only one local elasticsearch.
Data is in sync. But seems like both are trying to be master nodes.

How can I resolve this?


(Vilas Reddy Podduturi) #2

I know I can set in one of the configuration that it cannot be master node.
But then I can't bounce the elasticsearch as there might be missing master node.
Please suggest.

(Mark Walkom) #3

What does the rest of your config (minus comments) look like?

(Vilas Reddy Podduturi) #4

All other config is default except this:

path.conf: ${ELASTIC}/conf
path.data: ${ELASTIC}/data
path.logs: ${ELASTIC}/logs
path.plugins: ${ELASTIC}/plugins

(Imran Siddique) #5

did u try adding other nodes' name in unicast list and trying: https://www.elastic.co/guide/en/elasticsearch/reference/current/modules-discovery-zen.html#unicast

(Mark Walkom) #6

Big ++ on that last one.

Also take a look at https://www.elastic.co/guide/en/elasticsearch/guide/current/_important_configuration_changes.html

(Vilas Reddy Podduturi) #7

Thanks @mosiddi, it worked.

@warkolm, We have only two servers in the cluster, and I see the recommendation to use 3 servers and quorum of 2.
Currently minimum_master_nodes is 1. Can we continue using this setting?

(Mark Walkom) #8

You can but you are at greater risk of split brain.

(Imran Siddique) #9

agreed with Mark.

(Vilas Reddy Podduturi) #10

Yes. We see the risk. Planning to add another server to the production to avoid it.
Thanks for your inputs.

(system) #11