Get kubeconfig for custom cluster in Rancher 2.x

get_kubeconfig_custom_cluster_rancher2.sh

#!/bin/bash
# Usage: ./get_kubeconfig_custom_cluster_rancher2.sh cluster_name
# Needs to be run on the server running `rancher/rancher` container

# Check if jq exists
command -v jq >/dev/null 2>&1 || { echo "jq is not installed. Exiting." >&2; exit 1; }

# Check if clustername is given
if [ -z "$1" ]; then
    echo "Usage: $0 [clustername]"
    exit 1
fi

# Provide clustername as first argument
CLUSTERNAME=$1

# Retrieve Docker container ID of container running `rancher/rancher` image
CONTID=$(docker ps | grep -E "rancher/rancher:|rancher/rancher |rancher/rancher@|rancher_rancher" | awk '{ print $1 }')
echo "Container ID running Rancher is ${CONTID}"

# Validate that we are querying the correct etcd
if docker exec $CONTID kubectl get clusters.management.cattle.io > /dev/null; then
  echo "'kubectl get cluster' returns clusters available"
else
  echo "'kubectl get cluster' returns error, this should be run on the host running the 'rancher/rancher' container or embedded kubectl of the 'local' imported cluster"                                                                                                                   
  exit 1
fi

# Get clusters
CLUSTERINFO=$(docker exec $CONTID kubectl get clusters.management.cattle.io --no-headers --output=custom-columns=Name:.spec.displayName,Driver:.status.driver,ClusterID:.metadata.name)                                                                                                                          
echo "Clusters found:"
echo "${CLUSTERINFO}"

# Get clusterid from clustername
CLUSTERID=$(echo "${CLUSTERINFO}" | awk -v CLUSTERNAME=$CLUSTERNAME '$1==CLUSTERNAME { print $3 }')

if [[ -z $CLUSTERID ]]; then
  echo "No CLUSTERID could be retrieved for $CLUSTERNAME, make sure you entered the correct clustername"
  exit 1
fi

CLUSTERDRIVER=$(echo "${CLUSTERINFO}" | awk -v CLUSTERNAME=$CLUSTERNAME '$1==CLUSTERNAME { print $2 }')

if [[ $CLUSTERDRIVER != "rancherKubernetesEngine" ]]; then
  echo "Cluster ${CLUSTERNAME} is not a RKE built cluster, you can't retrieve the kubeconfig with this script"
  echo "If you have built your cluster using RKE and then imported, the 'kube_config_cluster.yml' file can be used"
  echo "If you have imported your own cluster, the kubeconfig was created when building your cluster"
  exit 1
fi

# Get kubeconfig for cluster ID and save it to `kubeconfig`
if docker exec $CONTID kubectl get secret c-$CLUSTERID -n cattle-system; then
  echo "Secret c-${CLUSTERID} found"
  docker exec $CONTID kubectl get secret c-$CLUSTERID -n cattle-system -o json | jq -r .data.cluster | base64 -d | jq -r .metadata.state > kubeconfig                                                                                                                                      
  if [[ -s kubeconfig ]]; then
    echo "Kubeconfig written to file 'kubeconfig'"
  else
    echo "Kubeconfig could not be retrieved"
  fi
else
  echo "Secret c-${CLUSTERID} could not be found"
  exit 1
fi

# Show kubeconfig
cat kubeconfig

if command -v kubectl >/dev/null 2>&1; then
    # Run kubectl against kubeconfig
    kubectl --kubeconfig kubeconfig get nodes
fi

oneliner-embedded-kubectl.sh

#!/bin/bash
# Needs to be run in embedded kubectl of Imported cluster called local, or using the RKE generated kubeconfig

# Check if jq exists
command -v jq >/dev/null 2>&1 || { echo "jq is not installed. Exiting." >&2; exit 1; }

# Check if clustername is given
if [ -z "$1" ]; then
    echo "Usage: $0 [clustername]"
    exit 1
fi

# Provide clustername as first argument
CLUSTERNAME=$1

# Validate that we are querying the correct etcd
if kubectl get clusters.management.cattle.io > /dev/null; then
  echo "'kubectl get cluster' returns clusters available"
else
  echo "'kubectl get cluster' returns error, this should be run in the embedded kubectl of the 'local' imported cluster"                                                                                                                                                                   
  exit 1
fi

# Get clusters
CLUSTERINFO=$(kubectl get clusters.management.cattle.io --no-headers --output=custom-columns=Name:.spec.displayName,Driver:.status.driver,ClusterID:.metadata.name)
echo "Clusters found:"
echo "${CLUSTERINFO}"

# Get clusterid from clustername
CLUSTERID=$(echo "${CLUSTERINFO}" | awk -v CLUSTERNAME=$CLUSTERNAME '$1==CLUSTERNAME { print $3 }')

if [[ -z $CLUSTERID ]]; then
  echo "No CLUSTERID could be retrieved for $CLUSTERNAME, make sure you entered the correct clustername"
  exit 1
fi

CLUSTERDRIVER=$(echo "${CLUSTERINFO}" | awk -v CLUSTERNAME=$CLUSTERNAME '$1==CLUSTERNAME { print $2 }')

if [[ $CLUSTERDRIVER != "rancherKubernetesEngine" ]]; then
  echo "Cluster ${CLUSTERNAME} is not a RKE built cluster, you can't retrieve the kubeconfig with this script"
  echo "If you have built your cluster using RKE and then imported, the 'kube_config_cluster.yml' file can be used"
  echo "If you have imported your own cluster, the kubeconfig was created when building your cluster"
  exit 1
fi

# Get kubeconfig for cluster ID and save it to `kubeconfig`
if kubectl get secret c-$CLUSTERID -n cattle-system; then
  echo "Secret c-${CLUSTERID} found"
  kubectl get secret c-$CLUSTERID -n cattle-system -o json | jq -r .data.cluster | base64 -d | jq -r .metadata.state > kubeconfig
  KUBECONTENT=$(cat kubeconfig)
  if [[ "${KUBECONTENT}" != "null" ]]; then
    echo "Kubeconfig written to file 'kubeconfig'"
  else
    echo "Kubeconfig could not be retrieved"
  fi
else
  echo "Secret c-${CLUSTERID} could not be found"
  echo "The cluster is not provisioned yet or is not in healthy state"
  exit 1
fi

$ kubectl get cluster
the server doesn't have a resource type "cluster"
$ kubectl get
You must specify the type of resource to get. Valid resource types include:

• all
• certificatesigningrequests (aka 'csr')
• clusterrolebindings
• clusterroles
• componentstatuses (aka 'cs')
• configmaps (aka 'cm')
• controllerrevisions
• cronjobs
• customresourcedefinition (aka 'crd')
• daemonsets (aka 'ds')
• deployments (aka 'deploy')
• endpoints (aka 'ep')
• events (aka 'ev')
• horizontalpodautoscalers (aka 'hpa')
• ingresses (aka 'ing')
• jobs
• limitranges (aka 'limits')
• namespaces (aka 'ns')
• networkpolicies (aka 'netpol')
• nodes (aka 'no')
• persistentvolumeclaims (aka 'pvc')
• persistentvolumes (aka 'pv')
• poddisruptionbudgets (aka 'pdb')
• podpreset
• pods (aka 'po')
• podsecuritypolicies (aka 'psp')
• podtemplates
• replicasets (aka 'rs')
• replicationcontrollers (aka 'rc')
• resourcequotas (aka 'quota')
• rolebindings
• roles
• secrets
• serviceaccounts (aka 'sa')
• services (aka 'svc')
• statefulsets (aka 'sts')
• storageclasses (aka 'sc')error: Required resource not specified.
  Use "kubectl explain " for a detailed description of that resource (e.g. kubectl explain pods).
  See 'kubectl get -h' for help and examples.

Has kubectl updated since this script, or am i using a different version? Cluster is not a resource

Sorry, I spoke too soon. I see now that kubectl is running inside the rancher docker container, and that command works fine. I had to modify the docker ps | grep "rancher/rancher" because my image name is rancher/rancher and doesn't match any of grep -E "rancher/rancher:|rancher/rancher@|rancher_rancher".

I am trying to get this work on our self hosted GitLab server but I had no success yet. I successfully managed to run your script and get generate kubeconfig, creating the CA certificate etc. I created the API key manually on my Rancher cluster but when I enter all the info in "Add Kubernetes cluster" in a GitLab project, I get:

Something went wrong while installing Helm Tiller
Can't start installation process.

Any ideas?

@tamasgal Did you figure it out? I suspect that if 'helm' already has a tiller service registered that you will encounter this error.

I got an issue when running the "oneliner-embedded-kubectl.sh":

base64: invalid option -- d

Updated to "-D" solved the issue.