Facing issues after upgrading EKS from v1.25 to v1.26

Question

After upgrading EKS cluster from v1.25 to v1.26 we are facing memory issues (oom error) in some of our applications. How to solve this issue. And also can we rollback the update to v1.25 again?

Answer

Once you upgrade a cluster, you can't downgrade to a previous version.

The only way to get to a previous version would be to deploy a new cluster. It may be worth in the future to instead of upgrading the cluster, create a 2nd cluster and move your pods across in a phased migration/upgrade in case you have any of these issues again.

https://docs.aws.amazon.com/eks/latest/userguide/update-cluster.html

Similar issue here https://learn.microsoft.com/en-us/troubleshoot/azure/azure-kubernetes/aks-memory-saturation-after-upgrade

Answer

You might look if you could you be impacted by the rollout of CgroupsV2. The standard AL2023-based EKS-optimized AMI uses Control Group version 2 (cgroup v2).   Old JAVA runtimes are not cgroupv2 aware and can not autodetect container memory limits ... leading to OOMs.   Just in case.

Facing issues after upgrading EKS from v1.25 to v1.26

Relevanter Inhalt