amazon-eks-ami
amazon-eks-ami copied to clipboard
awslabs
Packer configuration for building a custom EKS AMI
## Question: Is there an easy/simple way to feed a list of additional container images to the AMI builder or builder scripts? ### Use-case: Using something like Keda, we want...
**What happened**: kubelet is flapping between `NodeReady` and `NodeNotReady`, because of the following error: ``` "Skipping pod synchronization" err="PLEG is not healthy: pleg was last seen active 3m11.637412161s ago; threshold...
Kubernetes CI system uses specific versions of containerd and runc: - https://github.com/search?q=repo%3Akubernetes%2Ftest-infra+CONTAINERD_VERSION&type=code - https://github.com/search?q=repo%3Akubernetes%2Ftest-infra+RUNC_VERSION&type=code Very often folks in the kubernetes community use these versions that are specifically used in the...
When folks customize their AMI(s) it would be good to have some way to validate that they haven't broken any expectations of the maintainers of this repository ( with a...
**What happened**: The node indicated that it had plenty of free space and inodes, but attempts to create new files resulted in an `ENOSPC` error. ``` [root@ip-192-168-41-179 tmp]# df -ah...
**What would you like to be added**: I would like to add support for configuring a host and kubelet for use with a network proxy to the bootstrap.sh (or add...
Is there a way support for estargz could be added to the eks AMI? Especially for large images (~12GB) , e.g. for ML training, pulling a private image from the...
**Description of changes:** By default the OS+k8s components consumes at-least 2+Gi on any empty node. Having the default values of KubeReserved:ephemeral-storage set to **1Gi**, leads to disk-pressure for pods heavily...
**What would you like to be added**: A valid way to apply taints to nodes at launch **Why is this needed**: According to the k8s [kubelet reference docs](https://kubernetes.io/docs/reference/command-line-tools-reference/kubelet/), the `register-with-taints`...
**What happened**: Nodes went out of comission due to memory/cpu pressure from workloads. **What you expected to happen**: Nodes to reserve a certain amount of cpu/memory for OS and remain...
