Elastic Master nodes cluster heap usage

In our Elasticsearch cluster we have 3 master nodes. I've noticed that from time to time one of them is under a heavy load while the other two seem to be doing nothing. Is that normal for elastic master nodes or is that a sign of something bad?

It looks like this in Cerebro:

Without the column headings it's a little hard to be sure what these numbers mean. However, Elasticsearch does elect a single master-eligible node to be in charge of the cluster and this elected node does perform a lot more work than the other master-eligible ones, so this kind of mismatch seems reasonable to me.

Sadly, it is not the "master of masters" that is under heavy load.

That's how HEAP USAGE of that specific node looks in Kibana's monitoring:

Two other master nodes, including the "master of masters":

I think i found the problem. Garbage Collector on that "strange" node.
Had to correct JVM settings on that one node. Problem solved.

1 Like

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