Pending Pods: Limits in EKS

It is annoying to deploy our application in an EKS cluster finally, and have our pods remain in the eternal “pending” state.

Several reasons might cause a pod to be in a pending state; most of them are related to computing resource limits, but some others are related to IP address limits in our worker nodes.

However, reaching this kind of limit is a relatively straightforward problem to solve. You can find the right type of instance for your Kubernetes workloads by checking this handy document provided by AWS. To have the maximum number of pods per instance, and best of all, not only a few but all AWS EKS instances are available.

So, if you want to avoid some frustrations and headaches, that document will be a good starting point!

If your pods remain Pending due to node or network configuration issues—even after adjusting limits—you may need to access the EKS worker nodes manually for deeper diagnostics. We recommend using AWS Systems Manager Session Manager for secure, audited access to your instances without opening SSH ports.

Come back later for more Teratips about EKS.

Leandro Mansilla

DevOps Engineer

Teracloud