Elasticsearch 7.11 low heap size with bundled JDK

Hi all,
I am currently building a brand new cluster based on 7.11. In the docs under

it is said that

By default, Elasticsearch automatically sizes JVM heap based on a node’s roles and total memory. We recommend this default sizing for most production environments.

So my data nodes have 64 GB of RAM. There are defined as nodes.roles: data_warm.
GET /_cat/nodes?v=true
also shows me the role "w".
But the Bundled JDK just starts with a heapsize of 2GB and I would like to know, why so less? :slight_smile:

[2021-02-17T12:54:00,447][INFO ][o.e.n.Node               ] [el-dn-101] JVM home [/usr/share/elasticsearch/jdk], using bundled JDK [true]
[...]
[2021-02-17T12:54:02,877][INFO ][o.e.e.NodeEnvironment    ] [el-dn-101] heap size [2gb], compressed ordinary object pointers [true]

Cheers,
Marcus

Hello Marcus,

How did you install the ElasticSearch software and on which system are you running? Did you have another ElasticSearch instance running before?

This sounds as if ElasticSearch would have an explicit heap size declared(2GB was the default if I remember correctly).

This could be configured in an Elasticsearch config file under config/jvm.options

Best regards
Wolfram

Hi Wolfram,
it's a fresh installed Ubuntu 20.04.2 LTS. Installed via the Ansible Playbook from elastic.elasticsearch

Edit:
It seems that the role sets a default value, when you don't set es_heap_size .
I create an issue in github and comment out the

-Xms2g
-Xmx2g

in jvm.conf

1 Like