How to increase JVM heap size?

[desertSniper87]

How to increase the JVM heap size from 512m to 1G?

The official docs say this to pass as environment variable -Xms1g -Xmx1g: https://www.elastic.co/guide/en/elastic-stack-get-started/current/get-started-stack-docker.html#docker-enroll-nodes

[desertSniper87]

We can change the heap size by editing docker-compose.yml and passing ES_JAVA_OPTS environment variable to elasticsearch container.

  elasticsearch:
   ...
    environment:
+     - ES_JAVA_OPTS=-Xmx1g -Xms1g

Added a pull request to change the readme #46 (comment)

[peasead]

I think those are legacy docs. If you look you can see it's for version 8.2.

Looking here at 8.14 (the current version), it says

By default, Elasticsearch automatically sets the JVM heap size based on a node’s roles and total memory. Using the default sizing is recommended for most production environments.

Do you see something that recommends adjusting the heap size in 8.14?

[peasead]

Also, is this different than what's in the docker-compose.yml file?

elastic-container/docker-compose.yml

Lines 97 to 101 in 72e2879

	mem_limit: ${MEM_LIMIT}
	ulimits:
	memlock:
	soft: -1
	hard: -1

[FlorianHeigl]

@peasead definitely doesn't autotune the heap size. this toy instance only has one node added in fleet and as such performs well with that sizing of '1GB'. but I don't feel it's good for that. the mem_limit (AFAIK set via cgroups) and the ES_JAVA_OPTS heap size are different things. As this is the only instance running it should mean it has all applicable roles also, so any auto-tuning should result in a higher target value also.

I get the following states:

That's with 224GB RAM and nothing else consuming it.
I think the general rule is that over 32GB needs some garbage collector tuning, but if it were autotuning, i'd expect it to at least go up to 8GB or so so it has wiggle room.