EKS Support Resources
When your own troubleshooting runs out, the speed of the answer depends mostly on where you ask and what evidence you bring. This page lists the places worth knowing and how to use them.
Official documentation
| Resource | Use it for |
|---|---|
| EKS User Guide | How each feature works and is configured |
| EKS API Reference | Exact request and response shapes for automation |
| EKS Best Practices Guide | Opinionated guidance on security, reliability, networking, scaling, cost and upgrades — the most useful single document |
| EKS Workshop | Hands-on labs for almost every feature (source) |
| EKS FAQs | Pricing, limits and support-policy questions |
The best-practices guide is worth reading in full once, not only when something breaks.
AWS Support
A paid support plan is the only channel with a response-time commitment and the only one that can look at the managed control plane, which you cannot see into.
What AWS Support can do: inspect control-plane health and logs, confirm service-side issues, explain EKS and add-on behaviour, and escalate to the service team.
What it will not do: debug your application, or support heavily customised third-party components beyond best effort.
Writing a case that gets answered quickly
Include, in the first message:
- Cluster ARN and region, Kubernetes version, platform version.
- What you expected and what happened, with exact error text.
- When it started (UTC) and what changed around then.
- Scope: one pod, one node, one zone, or everything.
- Resource identifiers: instance IDs, node names, pod names and namespaces, add-on versions, request IDs from failed API calls.
- What you have already ruled out.
- Diagnostic bundles (below).
Choose severity honestly — “production system down” gets a fast response and should be reserved for that. For anything urgent, ask for a call or chat rather than waiting on email.
Diagnostic tooling
| Tool | What it produces |
|---|---|
EKS log collector script (in the awslabs/amazon-eks-ami repository) | A tarball of kubelet, containerd, CNI, iptables, kernel and system logs from a node — attach it to the case |
AWSSupport-TroubleshootEKSWorkerNode (Systems Manager automation) | An automated check of why a node will not join |
| Cluster insights | Upgrade-readiness and configuration findings |
| EKS observability dashboard | Control-plane metrics and health in the console |
| VPC Reachability Analyzer | Whether a network path exists between two ENIs |
aws eks describe-addon / describe-nodegroup | Health issues reported by the service itself |
Enable control plane logging before you need it; Support will ask for those logs and they cannot be produced retroactively.
Community
| Channel | Good for |
|---|---|
| AWS re:Post — EKS | Searchable Q&A; AWS staff answer there |
| Kubernetes Slack | #eks, plus channels for individual projects (#karpenter, #cilium) |
| AWS Developers Slack | General AWS discussion |
| Project GitHub issues | Bugs in a specific component — search before filing |
Community channels are best for “how do I” and “has anyone seen this” questions. They are not a place for account-specific details or anything urgent. Remove account IDs, ARNs and secrets from anything you paste.
Source repositories
Much of EKS is open source, and reading the code or the issue tracker often answers a question faster than waiting.
| Repository | Contains |
|---|---|
aws/amazon-vpc-cni-k8s | The VPC CNI |
awslabs/amazon-eks-ami | The node AMI build and nodeadm |
kubernetes-sigs/aws-load-balancer-controller | ALB and NLB integration |
kubernetes-sigs/aws-ebs-csi-driver, aws-efs-csi-driver | Storage drivers |
aws/karpenter-provider-aws | Karpenter for AWS |
aws-controllers-k8s/community | ACK |
aws/eks-charts | Helm charts for AWS components |
aws/eks-distro | The Kubernetes distribution EKS is built from |
Staying ahead of changes
- Roadmap: the public
aws/containers-roadmaprepository shows what is planned and lets you vote; the EKS Distro project board tracks its releases. - Version calendar: the Kubernetes release calendar in the user guide lists end-of-standard-support and end-of-extended-support dates for each version. Put them in your own calendar.
- AWS Health Dashboard: account-specific notices — scheduled maintenance, deprecations and required actions for your clusters. Route these events through EventBridge to the team’s channel.
- Release notes: platform versions, add-on versions and AMI releases each have their own changelog.
- What’s New feed for EKS feature launches.
Deadlines missed here turn into forced upgrades — see the upgrade process.
Before asking anyone
A short checklist that resolves a surprising share of problems:
- Read the events:
kubectl describeandkubectl get events --sort-by=.lastTimestamp - Checked common EKS issues for the symptom
- Add-ons are on supported, current versions
- Looked at what changed in the last day: deployments, IAM, security groups, node AMIs
- Searched the component’s GitHub issues for the exact error string
- Reproduced it in the smallest possible case