Elasticsearch – Best practices to reboot Database on Linux

Solved663 viewsdatabase Elastic Elasticsearch Linux OS restart
1
0 Comments

Hi Dojo,

Are there any requirements or recommendations to do a Elasticsearch Database rebooting? I have some recommendations about Cassandra but I have no information about Elastic.

Note that we're running Elastic on Linux OS.

Daniel Martinez [DevOps Enabler] Selected answer as best 29th May 2023

1 Answer

6
1.02K 1 Comment

Hi Daniel,

It's best to disable shard allocation before doing any reboots

(it's going to mangle the json I'm going to put below)

Send the following to http://<cluster node>:9200/_cluster/settings

{
  "transient" : {
    "cluster.routing.allocation.enable": "none",
    "cluster.routing.rebalance.enable" : "none"
  }
}

You can then reboot each node in turn as required.

Replace
"none" with null (note no quotes on null)
to renable

Once your nodes have all been restarted, send a retry shard allocation HTTP POST request
http://<cluster node>:9200/_cluster/reroute?retry_failed=true

You'll see after a few minutes any indicies with yellow status turn green

Hope that helps
Chris

Daniel Martinez [DevOps Enabler] Selected answer as best 29th May 2023