Constant GC warnings with small indexes

Hello everyone,

My team is getting constant GC overhead warnings on our production Elasticsearch cluster for our Liferay application, at least 5 times each minute. This occurs any time, not just when reindexing.
"message": "[gc][665978] overhead, spent [986ms] collecting in the last [1s]",

Here are our stats:
OS: Amazon linux 2
Ram: 32 gb
Heap: 20 gb
Disk: 90gb
Java version: 8 ES
Elasticsearch version: 7.4.2

I have made a gist with the output of the cluster stats API.

We are very confused because our indexes are not very large and we avoid expensive queries.

We were wondering if anyone had any tips or thoughts on this.

Thanks very much!

Welcome to our community! :smiley:

7.4 is EOL, please upgrade.

That doesn't look like the output I'd expect to see from that API. There's no information about the nodes, JVM or OS.

Thanks for your response! We've been working on updating our test environment to the latest ES version. It doesn't seem to have helped this problem but it's definitely something we worth doing, we will get that going on prod soon.

Sorry about the gist, not sure what happened there, here is an accurate version if anyone has time to take a peek, they are from prod so the version upgrade won't be there yet.

Thanks again!

Is there anything else running on the host where Elasticsearch is running? Do you have swap enabled?

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