kubectl config current-context printed error: current-context is not set on 23 October 2026, so there was no Amazon EKS cluster version to move from 1.34 to 1.35. Docker Engine 29.8.2 on that OrbStack host had Swarm inactive, and its 67 image names contained no kube, eks, k3s, kind, or k8s string.
Do not change an EKS control-plane version until that command prints the context you mean to change. This does not apply when the command already prints that context.
By Sudipta Dhara
Tested 2026-10-23
AI disclosure: This page was drafted with AI assistance and checked against the commands and the pages linked below on 23 October 2026.
Who this is for: DevOps engineers · Intermediate · a host you believe is an EKS node · You’ll need: kubectl and docker on that host
Shareable fact: On the Amazon EKS version calendar read on 23 October 2026, Kubernetes 1.34 reaches end of standard support on 2 December 2026 and end of extended support on 2 December 2027. Caveat: those dates do not create a cluster, and the same day’s kubectl config current-context still printed error: current-context is not set.
Did the host have a cluster context?
We read the context before any version change. The command and the output on 23 October 2026 were:
kubectl config current-context
error: current-context is not set
aws, eksctl, kind, and k3s were not on PATH. No AWS API call was made. There was no cluster version to print, so the upgrade steps in the user guide had nowhere to run.
What did Docker show on 23 October 2026?
The same host reported Docker Engine 29.8.2. docker info printed OS=OrbStack and Swarm=inactive. docker images printed 67 repository:tag lines. None of those lines contained kube, eks, k3s, kind, or k8s. That is a Docker engine with no Kubernetes image name in the local list. It is not an EKS control plane, and it is not a record of your cluster.
If this host is the box you meant to upgrade, stop. A full disk, a memory cap, or an Ubuntu upgrade is a different job. Those checks are in Docker disk full, Docker memory limits, and Ubuntu 24.04 to 26.04 on a Docker host. This read did not measure disk use or memory.
When does the 2 December 2026 date apply?
Read the calendar only after kubectl config current-context prints a context. On 23 October 2026 the Amazon EKS version calendar listed 1.37, 1.36, 1.35, and 1.34 in standard support. The 1.34 row says upstream release 27 August 2025, Amazon EKS release 2 October 2025, end of standard support 2 December 2026, and end of extended support 2 December 2027. The same page says a minor version stays in standard support for 14 months after release, then extended support for 12 months at an additional cost per cluster hour. Extended support is enabled by default. This draft states no price.
| Kubernetes version | End of standard support | End of extended support |
|---|---|---|
| 1.34 | 2 December 2026 | 2 December 2027 |
| 1.35 | 27 March 2027 | 27 March 2028 |
Caveat: two rows from the calendar page on 23 October 2026. They do not name your context.
The update guide, also read on 23 October 2026, says you can update only one minor version at a time, and that you can roll back to the previous version within 7 days. The Kubernetes 1.35 notes on EKS, read the same day, say the kubelet refuses to start by default on cgroup v1 nodes, AL2023 uses cgroup v2 by default, the --pod-infra-container-image flag is removed, and Fargate continues to use cgroup v1. On 23 October 2026 the Kubernetes v1.35.0 release page returned HTTP 200 and the HTML contained Release Kubernetes v1.35.0. That fetch did not include a changelog body, so no cgroup line is quoted from GitHub. On 23 October 2026 the version skew policy said kubectl is supported within one minor version, older or newer, of kube-apiserver. Its example is kube-apiserver at 1.37 and kubectl at 1.38, 1.37, or 1.36.
Who is this not for?
This is not for a cluster whose kubectl config current-context already prints the name you intend to change. It is not an EKS build offer. The services page read on 23 October 2026 had no EKS heading.
What are the alternatives?
The alternative to starting the upgrade on a host with no context is to run kubectl config current-context on the host you think is the cluster and to stop when it errors. A second alternative is to stay on the current minor until that command succeeds, then follow the one-minor step in the update guide. A Docker disk, memory, or Ubuntu problem stays on the three guides linked above.
What are the cons?
The calendar date can still arrive while this host has no context. Stopping here does not move a real 1.34 control plane, and it does not test cgroup v1, Fargate, or a custom AMI flag. Extended support has a cost per cluster hour that this draft does not quote. The GitHub release HTML did not include the changelog, so the cgroup lines rest on the EKS 1.35 notes, not on a kubelet log from this host.
What broke, and what fixed it?
The 1.34 to 1.35 checklist assumed a context. The first command printed error: current-context is not set, and the image list had no Kubernetes name. Trying the user-guide steps next would have described a cluster this host does not have. Stopping at the context line fixed that. The calendar stays a dated source for the day a real 1.34 context still needs a one-minor move to 1.35.
FAQ
Did kubectl have a context on 23 October 2026?
No. kubectl config current-context printed error: current-context is not set.
Does the 2 December 2026 date mean you should upgrade today?
Only after kubectl config current-context prints the cluster you intend to change. The calendar date is not a context.
Raw readings
These checks do not add a context. On 23 October 2026 aws was missing, eksctl was missing, kind was missing, and k3s was missing. AWS_ACCESS_KEY_ID was empty. docker info also printed Name=orbstack.
Next step
On the host you mean to upgrade, run kubectl config current-context. If it prints error: current-context is not set, stop. If it prints a name, open the version calendar and confirm that name is the 1.34 cluster before you change one minor version.
Talk to ProDevs about a Docker host that fails a command: https://prodevs.in/services/
