Quickstart for distributed connector in kubernetes, connect-0 OOMKilled

I’m following the instructions here - Confluent for Kubernetes Quick Start | Confluent Documentation. When I get 4.3, all the nodes are ok except…

connect-0 0/1 CrashLoopBackOff 11 (43s ago) 50m

Last State:     Terminated
  Reason:       OOMKilled

did you check the nodes of your underlying K8s cluster?

I was running on an EC2 instance that had 32G memory. Switching to a 72GB EC2 instance it no longer gets OOM. I did not see that in the prerequisites. I started at the URL in the README.md

Confluent for Kubernetes is a complete, declarative API for deploying and operating Confluent Platform as a cloud-native system on Kubernetes.

Documentation for this is provided at Deploy and Manage Confluent Platform Using Confluent for Kubernetes | Confluent Documentation .

This package includes the following:

  • The Helm Chart to deploy Confluent for Kubernetes

And then the QuickStart link.

I think 32GB should have been enough at least for testing.
looks strange a bit

though for prod deployments there are some recommendations