Helm test pods OOMKilled

devsuv · February 23, 2023, 10:04am

I have deployed many application pods in multi worker node Kubernates setup. When I run helm test all the helm test pods present in different worker are in OOMKilled state?
Any suggetion how to fix or debug this?

mmkmou · February 23, 2023, 1:56pm

Hi,

OOMKilled state means pod used more memory than allowed. Review the request memory and CPU and deploy again

Regard

B-Gopi · December 7, 2023, 5:15pm

Hi,
Can we use memoryQoS to handle OOMError by throttling memory? if so, can you help me how to do it in AWS EKS?

Alistair_KodeKloud · December 27, 2023, 3:02pm

You can’t “throttle” memory. If a pod tries to use more than its configured limit, it will be OOMKilled. This is how the underlying Cgroup works. Cgroup is a feature of Linux itself, not Kubernetes. Kubelet simply leverages this operating system functionality when creating containers.