Kubernetes cluster unreachable

Helm reads the same kubeconfig as kubectl, and right now it points at a cluster that is not there - missing KUBECONFIG, wrong context, or the cluster is down. Run kubectl cluster-info: if kubectl fails too, the problem is the kubeconfig, not helm. Fix the config (check $KUBECONFIG, kubectl config current-context, re-fetch credentials) and helm works.

The error

Error: Kubernetes cluster unreachable: Get "YOUR_HOST:6443/version": dial tcp YOUR_HOST:6443: connect: connection refused

What to do

1. Check whether kubectl agrees:

kubectl cluster-info

Expected: If this fails too, the problem is the kubeconfig, not helm.

1. Inspect the config:

echo $KUBECONFIG
kubectl config current-context

Expected: Identifies a missing file or the wrong context.

1. Fix it: kubectl config use-context [correct-context] or re-fetch credentials from your provider.

Expected: Context switched / config written.

1. Retry:

helm version
helm install [release] [chart]

Expected: No unreachable error.

When this applies

• the exact Kubernetes cluster unreachable message
• fresh machines, new shells, CI runners without kubeconfig
• contexts pointing at deleted clusters

When it does NOT apply

• RBAC forbidden errors (connected, but denied)
• chart template or value errors

Works with

helm 3.x; any cluster

Error: INSTALLATION FAILED: Kubernetes cluster unreachable: the server could not find the requested resource

Config points at something, but the API answers oddly. Usually a stale or partial kubeconfig - regenerate it.

Why it happens

Helm is a client of the Kubernetes API like kubectl is; it autodetects the same kubeconfig. An unreachable cluster for helm is an unreachable cluster for kubectl - diagnose at the kubectl layer.

Edge cases

• sudo drops your KUBECONFIG and home: either run helm as your user or pass --kubeconfig explicitly.
• k3s/microk8s keep their kubeconfig outside ~/.kube - export KUBECONFIG to it or copy it into place.

Resolved from

oneuptime helm troubleshooting guide - https://oneuptime.com/blog/post/2026-01-17-helm-troubleshooting-common-errors/view