Hi, Since Java 10 and openJdk 8u212 - there was a container-awareness feature added... Basically it makes JVM recognise resource allocation of the container it is running in (if that's the case).
We ran into an issue with ElasticSearch - since ES is utilizing this feature and creates internal threads per the resources available, whatever it thinks is available based on the info JVM gets back to it. When run in a Docker container, with no special setting for CPUs or RAM, by default the number of CPUs available to the process is returned as 1... even though there are many more cores available on the node. We were not doing this setting of resource for the Docker container since we are running one container per node, via Kubernetis, and assume that all resources of the node are available to the container and whatever it is running... Not the case with ES :) So, the question is: does Kafka also rely on the resources reported as available by the JVM or not? I search and googled - but did not find any specific info... thank you! Marina Sent with [ProtonMail](https://protonmail.com) Secure Email.
