Helm 'Error: Kubernetes cluster unreachable: Get "http://localhost:8080/version": dial tcp ... connection refused' (or 'the server has asked for the client to provide credentials'). How do I point Helm at the right cluster?

Cause: Helm talks to Kubernetes using the same kubeconfig as kubectl. 'Kubernetes cluster unreachable' with localhost:8080 means Helm found no usable kubeconfig, so it fell back to the default local address: KUBECONFIG not set (or set to a missing file), no ~/.kube/config for the user running Helm (sudo, CI, a container), or a kubeconfig with no current context. Credential errors mean the kubeconfig was found but its credentials are expired or invalid (cloud tokens, exec plugins like aws, gke-gcloud-auth-plugin or kubelogin not installed). Fix: 1. Check that kubectl works with the same user and environment: kubectl config current-context and kubectl get nodes. If kubectl fails, fix that first; Helm will then work too. 2. Point to the right file: export KUBECONFIG=/path/to/config (or pass --kubeconfig to helm), and select the context with kubectl config use-context <name> or helm --kube-context <name>. 3. Don't run helm with sudo unless root has a kubeconfig; root uses /root/.kube/config, not yours. 4. Refresh cloud credentials: aws eks update-kubeconfig --name <cluster>, gcloud container clusters get-credentials <cluster>, or az aks get-credentials, and install the auth plugin the kubeconfig's exec section names. 5. In CI or containers, mount or write the kubeconfig (from a secret) and set KUBECONFIG explicitly in the job's environment. 6. On k3s, the kubeconfig lives at /etc/rancher/k3s/k3s.yaml; copy it or set KUBECONFIG to it with suitable permissions. Verify: kubectl get nodes and helm list -A both succeed against the intended cluster. Sources: https://helm.sh/docs/faq/troubleshooting/ https://kubernetes.io/docs/concepts/configuration/organize-cluster-access-kubeconfig/