Start and Stop order of ELK nodes

I have the following setup.

Elasticsearch master nodes 3
Elasticsearch hot data nodes 2
Elasticsearch warm data nodes 2
Elasticsearch co-ordinating nodes and kafka nodes 2
Logstash 2 nodes
kafka 3 nodes
zookeeper 2 nodes.

What should be the start and stop order?

I'd do:

  • Elasticsearch master nodes 3
  • Elasticsearch hot data nodes 2
  • Elasticsearch warm data nodes 2
  • Elasticsearch co-ordinating nodes

For the rest, I'm not sure as I don't know what you are doing but I guess that Logstash should be started at the end.

Thank you. Just to make sure the master node shouldn't start with the shards related balancing activities (routing/allocation/creation ..etc ) till all nodes comes up (may be a delay due to network latency) , should we configure specific elasticsearch configuration parameters on master nodes?

You can disable rebalancing while restarting: cluster.routing.rebalance.enable

https://www.elastic.co/guide/en/elasticsearch/reference/current/shards-allocation.html#_shard_rebalancing_settings

Thanks. Whether this change has to be done only on master nodes? And once all nodes have restarted may be we need to set it back to all from none, through api?

It's better to do that only through the API.
Before stopping the cluster that is.

In case of DR there is no way we can use api. so while recovering I think manually the configuration file for master nodes to be updated?

I think that master nodes must have this set. I don't believe it's needed for data only nodes

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.