Skip to main content
Version: 16.x

Teleport GKE Auto-Discovery

The Teleport Discovery Service can automatically register your Google Kubernetes Engine (GKE) clusters with Teleport. With Teleport Kubernetes Discovery, you can configure the Teleport Kubernetes Service and Discovery Service once, then create GKE clusters without needing to register them with Teleport after each creation.

In this guide, we will show you how to get started with Teleport Kubernetes Discovery for GKE.

Overview​

Teleport cluster auto-discovery involves two components:

  1. The Teleport Discovery Service that watches for new clusters or changes to previously discovered clusters. It dynamically registers each discovered cluster as a kube_cluster resource in your Teleport cluster. It does not need connectivity to the clusters it discovers.
  2. The Teleport Kubernetes Service that monitors the dynamic kube_cluster resources registered by the Discovery Service. It proxies communications between users and the cluster.
tip

This guide presents the Discovery Service and Kubernetes Service running in the same process, however both can run independently and on different machines.

For example, you can run an instance of the Kubernetes Service in the same private network as the clusters you want to register with your Teleport cluster, and an instance of the Discovery Service in any network you wish.

Prerequisites​

  • A running Teleport cluster version 16.4.8 or above. If you want to get started with Teleport, sign up for a free trial or set up a demo environment.

  • The tctl admin tool and tsh client tool.

    Visit Installation for instructions on downloading tctl and tsh.

  • A Google Cloud account with permissions to create GKE clusters, IAM roles, and service accounts.
  • The gcloud CLI tool. Follow the Google Cloud documentation page to install and authenticate to gcloud.
  • One or more GKE clusters running. Your Kubernetes user must have permissions to create ClusterRole and ClusterRoleBinding resources in your clusters.
  • A Linux host where you will run the Teleport Discovery and Kubernetes services. You can run this host on any cloud provider or even use a local machine.
  • To check that you can connect to your Teleport cluster, sign in with tsh login, then verify that you can run tctl commands using your current credentials. For example: 
    $ tsh login --proxy=teleport.example.com --user=email@example.com
    $ tctl status
    # Cluster  teleport.example.com
    # Version  16.4.8
    # CA pin   sha256:abdc1245efgh5678abdc1245efgh5678abdc1245efgh5678abdc1245efgh5678
    If you can connect to the cluster and run the tctl status command, you can use your current credentials to run subsequent tctl commands from your workstation. If you host your own Teleport cluster, you can also run tctl commands on the computer that hosts the Teleport Auth Service for full permissions.

Step 1/3. Obtain Google Cloud credentials​

The Teleport Discovery Service and Kubernetes Service use a Google Cloud service account to discover GKE clusters and manage access from Teleport users. In this step, you will create a service account and download a credentials file for the Teleport Discovery Service.

Create an IAM role for the Discovery Service​

The Teleport Discovery Service needs permissions to retrieve GKE clusters associated with your Google Cloud project.

To grant these permissions, create a file called GKEKubernetesAutoDisc.yaml with the following content:

title: GKE Cluster Discoverer
description: "Get and list GKE clusters"
stage: GA
includedPermissions:
- container.clusters.get
- container.clusters.list

Create the role, assigning the --project flag to the name of your Google Cloud project:

$ gcloud iam roles create GKEKubernetesAutoDisc \
  --project=google-cloud-project \
  --file=GKEKubernetesAutoDisc.yaml

Create an IAM role for the Kubernetes Service​

The Teleport Kubernetes Service needs Google Cloud IAM permissions in order to forward user traffic to your GKE clusters.

Create a file called GKEAccessManager.yaml with the following content:

title: GKE Cluster Access Manager
description: "Manage access to GKE clusters"
stage: GA
includedPermissions:
- container.clusters.get
- container.clusters.impersonate
- container.pods.get
- container.selfSubjectAccessReviews.create
- container.selfSubjectRulesReviews.create

Create the role, assigning the --project flag to the name of your Google Cloud project. If you receive a prompt indicating that certain permissions are in TESTING, enter y:

$ gcloud iam roles create GKEAccessManager \
  --project=google-cloud-project \
  --file=GKEAccessManager.yaml

Create a service account​

Now that you have declared roles for the Discovery Service and Kubernetes Service, create a service account so you can assign these roles.

Run the following command to create a service account called teleport-discovery-kubernetes:

$ gcloud iam service-accounts create teleport-discovery-kubernetes \
  --description="Teleport Discovery Service and Kubernetes Service" \
  --display-name="teleport-discovery-kubernetes"

Grant the roles you defined earlier to your service account, assigning PROJECT_ID to the name of your Google Cloud project:

$ PROJECT_ID=google-cloud-project
$ gcloud projects add-iam-policy-binding ${PROJECT_ID?} \
   --member="serviceAccount:teleport-discovery-kubernetes@${PROJECT_ID?}.iam.gserviceaccount.com" \
   --role="projects/${PROJECT_ID?}/roles/GKEKubernetesAutoDisc"
$ gcloud projects add-iam-policy-binding ${PROJECT_ID?} \
   --member="serviceAccount:teleport-discovery-kubernetes@${PROJECT_ID?}.iam.gserviceaccount.com" \
   --role="projects/${PROJECT_ID?}/roles/GKEAccessManager"
Deploying the Kubernetes Service and Discovery Service separately?

Create a service account for each service:

$ gcloud iam service-accounts create teleport-discovery-service \
  --description="Teleport Discovery Service" \
  --display-name="teleport-discovery-service"
$ gcloud iam service-accounts create teleport-kubernetes-service \
  --description="Teleport Kubernetes Service" \
  --display-name="teleport-kubernetes-service"

Grant the roles you defined earlier to your service account, assigning PROJECT_ID to the name of your Google Cloud project:

$ PROJECT_ID=google-cloud-project
$ gcloud projects add-iam-policy-binding ${PROJECT_ID?} \
   --member="serviceAccount:teleport-discovery-service@${PROJECT_ID?}.iam.gserviceaccount.com" \
   --role="projects/${PROJECT_ID?}/roles/GKEKubernetesAutoDisc"
$ gcloud projects add-iam-policy-binding ${PROJECT_ID?} \
   --member="serviceAccount:teleport-kubernetes-service@${PROJECT_ID?}.iam.gserviceaccount.com" \
   --role="projects/${PROJECT_ID?}/roles/GKEAccessManager"

Retrieve credentials for your Teleport services​

Now that you have created a Google Cloud service account and attached roles to it, associate your service account with the Teleport Kubernetes Service and Discovery Service.

The process is different depending on whether you are deploying the Teleport Kubernetes Service and Discovery Service on Google Cloud or some other way (e.g., via Amazon EC2 or on a local network).

Stop your VM so you can attach your service account to it:

$ gcloud compute instances stop vm-name --zone=google-cloud-region

Attach your service account to the instance, assigning the name of your VM to vm-name and the name of your Google Cloud region to google-cloud-region:

$ gcloud compute instances set-service-account vm-name \
   --service-account teleport-discovery-kubernetes@${PROJECT_ID?}.iam.gserviceaccount.com \
   --zone google-cloud-region \
   --scopes=cloud-platform
Running the Kubernetes and Discovery Services separately?

Stop each VM you plan to use to run the Teleport Kubernetes Service and Discovery Service.

Attach the teleport-kubernetes-service service account to the VM running the Kubernetes Service:

$ gcloud compute instances set-service-account ${VM1_NAME?} \
   --service-account teleport-kubernetes-service@${PROJECT_ID?}.iam.gserviceaccount.com \
   --zone google-cloud-region \
   --scopes=cloud-platform

Attach the teleport-discovery-service service account to the VM running the Discovery Service:

$ gcloud compute instances set-service-account ${VM2_NAME?} \
   --service-account teleport-discovery-service@${PROJECT_ID?}.iam.gserviceaccount.com \
   --zone google-cloud-region \
   --scopes=cloud-platform
warning

You must use the scopes flag in the gcloud compute instances set-service-account command. Otherwise, your Google Cloud VM will fail to obtain the required authorization to access the GKE API.

Once you have attached the service account, restart your VM:

$ gcloud compute instances start vm-name --zone google-cloud-region

Step 2/3. Configure Teleport to discover GKE clusters​

Now that you have created a service account that can discover GKE clusters and a cluster role that can manage access, configure the Teleport Discovery Service to detect GKE clusters and the Kubernetes Service to proxy user traffic.

Install Teleport​

Install Teleport on the host you are using to run the Kubernetes Service and Discovery Service:

Install Teleport on your Linux server:

  1. Assign edition to one of the following, depending on your Teleport edition:

    EditionValue
    Teleport Enterprise Cloudcloud
    Teleport Enterprise (Self-Hosted)enterprise
    Teleport Community Editionoss

  2. Get the version of Teleport to install. If you have automatic agent updates enabled in your cluster, query the latest Teleport version that is compatible with the updater:

    $ TELEPORT_DOMAIN=example.teleport.com
    $ TELEPORT_VERSION="$(curl https://$TELEPORT_DOMAIN/v1/webapi/automaticupgrades/channel/default/version | sed 's/v//')"

    Otherwise, get the version of your Teleport cluster:

    $ TELEPORT_DOMAIN=example.teleport.com
    $ TELEPORT_VERSION="$(curl https://$TELEPORT_DOMAIN/v1/webapi/ping | jq -r '.server_version')"

  3. Install Teleport on your Linux server:

    $ curl https://cdn.teleport.dev/install-v16.4.8.sh | bash -s ${TELEPORT_VERSION} edition

    The installation script detects the package manager on your Linux server and uses it to install Teleport binaries. To customize your installation, learn about the Teleport package repositories in the installation guide.

Create a join token​

The Teleport Discovery Service and Kubernetes Service require an authentication token in order to to join the cluster. Generate one by running the following tctl command:

$ tctl tokens add --type=discovery,kube --format=text
abcd123-insecure-do-not-use-this

Copy the token (e.g., abcd123-insecure-do-not-use-this above) and save the token in /tmp/token on the machine that will run the Discovery Service and Kubernetes Service, for example:

$ echo abcd123-insecure-do-not-use-this | sudo tee /tmp/token
# abcd123-insecure-do-not-use-this
Running the Kubernetes and Discovery Services separately?

Generate separate tokens for the Kubernetes Service and Discovery Service by running the following tctl commands:

$ tctl tokens add --type=discovery --format=text
# efgh456-insecure-do-not-use-this
$ tctl tokens add --type=kube --format=text
# ijkl789-insecure-do-not-use-this

Copy each token (e.g., efgh456-insecure-do-not-use-this and ijkl789-insecure-do-not-use-this above) and save it in /tmp/token on the machine that will run the appropriate service.

Configure the Kubernetes Service and Discovery Service​

On the host running the Kubernetes Service and Discovery Service, create a Teleport configuration file with the following content at /etc/teleport.yaml:

warning

Discovery Service exposes a configuration parameter - discovery_service.discovery_group - that allows you to group discovered resources into different sets. This parameter is used to prevent Discovery Agents watching different sets of cloud resources from colliding against each other and deleting resources created by another services.

When running multiple Discovery Services, you must ensure that each service is configured with the same discovery_group value if they are watching the same cloud resources or a different value if they are watching different cloud resources.

It is possible to run a mix of configurations in the same Teleport cluster meaning that some Discovery Services can be configured to watch the same cloud resources while others watch different resources. As an example, a 4-agent high availability configuration analyzing data from two different cloud accounts would run with the following configuration.

  • 2 Discovery Services configured with discovery_group: "prod" polling data from Production account.
  • 2 Discovery Services configured with discovery_group: "staging" polling data from Staging account.
version: v3
teleport:
  join_params:
    token_name: "/tmp/token"
    method: token
  proxy_server: "teleport.example.com:443"
auth_service:
  enabled: off
proxy_service:
  enabled: off
ssh_service:
  enabled: off
discovery_service:
  enabled: "yes"
  discovery_group: "gke-myproject"
  gcp:
    - types: ["gke"]
      locations: ["*"]
      project_ids: ["myproject"] # replace with my project ID
      tags:
        "*" : "*"
kubernetes_service:
  enabled: "yes"
  resources:
  - labels:
      "*": "*"
Running the Kubernetes Service and Discovery Service on separate hosts?

Follow the instructions in this section with two configuration files. The configuration file you will save at /etc/teleport.yaml on the Kubernetes Service host will include the following:

version: v3
teleport:
  join_params:
    token_name: "/tmp/token"
    method: token
  proxy_server: teleport.example.com:443    
auth_service:
  enabled: off
proxy_service:
  enabled: off
ssh_service:
  enabled: off
kubernetes_service:
  enabled: "yes"
  resources:
  - labels:
      "*": "*"

On the Discovery Service host, the file will include the following:

version: v3
teleport:
  join_params:
    token_name: "/tmp/token"
    method: token
  proxy_server: teleport.example.com:443    
auth_service:
  enabled: off
proxy_service:
  enabled: off
ssh_service:
  enabled: off
discovery_service:
  enabled: "yes"
  discovery_group: "gke-myproject"
  gcp:
    - types: ["gke"]
      locations: ["*"]
      project_ids: ["myproject"] # replace with my project ID
      tags:
        "*" : "*"

Edit this configuration for your environment as explained below.

proxy_server​

Replace teleport.example.com:443 with the host and port of your Teleport Proxy Service (e.g., mytenant.teleport.sh:443 for a Teleport Cloud tenant).

discovery_service.gcp​

Each item in discovery_service.gcp is a matcher for Kubernetes clusters running on GKE. The Discovery Service periodically executes a request to the Google Cloud API based on each matcher to list GKE clusters. In this case, we have declared a single matcher.

Each matcher searches for clusters that match all properties of the matcher, i.e., that belong to the specified locations and projects and have the specified tags. The Discovery Service registers GKE clusters that match any configured matcher.

This means that if you declare the following two matchers, the Discovery Service will register clusters in project myproj-dev running in us-east1, as well as clusters in project myproj-prod running in us-east2, but not clusters in myproj-dev running in us-east2:

discovery_service:
  enabled: "yes"
  discovery_group: "gke-myproject"
  gcp:
    - types: ["gke"]
      locations: ["us-east1"]
      project_ids: ["myproj-dev"]
      tags:
        "*" : "*"
    - types: ["gke"]
      locations: ["us-east2"]
      project_ids: ["myproj-prod"]
      tags:
        "*" : "*"

discovery_service.gcp[0].types​

Each matcher's types field must be set to an array with a single string value, gke.

discovery_service.gcp[0].project_ids​

In your matcher, replace myproject with the ID of your Google Cloud project.

Ensure that the project_ids field follows these rules:

  • It must include at least one value.
  • It must not combine the wildcard character (*) with other values.
Examples of valid configurations​
  • ["p1", "p2"]
  • ["*"]
  • ["p1"]
Example of an invalid configuration​
  • ["p1", "*"]

discovery_service.gcp[0].locations​

Each matcher's locations field contains an array of Google Cloud region or zone names that the matcher will search for GKE clusters. The wildcard character, *, configures the matcher to search all locations.

discovery_service.gcp[0].tags​

Like locations, tags consists of a map where each key is a string that represents the key of a tag, and each value is either a single string or an array of strings, representing one tag value or a list of tag values.

A wildcard key or value matches any tag key or value in your Google Cloud account. If you include another value, the matcher will match all GKE clusters with the provided tag.

Start the Kubernetes Service and Discovery Service​

On the host where you will run the Kubernetes Service, execute the following command, depending on:

  • Whether you installed Teleport using a package manager or via a TAR archive
  • Whether you are running the Discovery and Kubernetes Service on Google Cloud or another platform

On the host where you will run the Teleport Kubernetes Service and Discovery Service, start the Teleport service:

$ sudo systemctl start teleport

Step 3/3. Connect to your GKE cluster​

Allow access to your Kubernetes cluster​

Ensure that you are in the correct Kubernetes context for the cluster you would like to enable access to:

$ kubectl config current-context
Using the wrong context?

Retrieve all available contexts:

$ kubectl config get-contexts

Switch to your context, replacing CONTEXT_NAME with the name of your chosen context:

$ kubectl config use-context CONTEXT_NAME
Switched to context CONTEXT_NAME

To authenticate to a Kubernetes cluster via Teleport, your Teleport user's roles must allow access as at least one Kubernetes user or group.

  1. Retrieve a list of your current user's Teleport roles. The example below requires the jq utility for parsing JSON:

    $ CURRENT_ROLES=$(tsh status -f json | jq -r '.active.roles | join ("\n")')

  2. Retrieve the Kubernetes groups your roles allow you to access:

    $ echo "$CURRENT_ROLES" | xargs -I{} tctl get roles/{} --format json | \
      jq '.[0].spec.allow.kubernetes_groups[]?'

  3. Retrieve the Kubernetes users your roles allow you to access:

    $ echo "$CURRENT_ROLES" | xargs -I{} tctl get roles/{} --format json | \
      jq '.[0].spec.allow.kubernetes_users[]?'

  4. If the output of one of the previous two commands is non-empty, your user can access at least one Kubernetes user or group, so you can proceed to the next step.

  5. If both lists are empty, create a Teleport role for the purpose of this guide that can view Kubernetes resources in your cluster.

    Create a file called kube-access.yaml with the following content:

    kind: role
    metadata:
      name: kube-access
    version: v7
    spec:
      allow:
        kubernetes_labels:
          '*': '*'
        kubernetes_resources:
          - kind: '*'
            namespace: '*'
            name: '*'
            verbs: ['*']
        kubernetes_groups:
        - viewers
      deny: {}

  6. Apply your changes:

    $ tctl create -f kube-access.yaml

  7. Assign the kube-access role to your Teleport user by running the appropriate commands for your authentication provider:

    1. Retrieve your local user's roles as a comma-separated list:

      $ ROLES=$(tsh status -f json | jq -r '.active.roles | join(",")')

    2. Edit your local user to add the new role:

      $ tctl users update $(tsh status -f json | jq -r '.active.username') \
        --set-roles "${ROLES?},kube-access"

    3. Sign out of the Teleport cluster and sign in again to assume the new role.

  8. Configure the viewers group in your Kubernetes cluster to have the built-in view ClusterRole. When your Teleport user assumes the kube-access role and sends requests to the Kubernetes API server, the Teleport Kubernetes Service impersonates the viewers group and proxies the requests.

    Create a file called viewers-bind.yaml with the following contents, binding the built-in view ClusterRole with the viewers group you enabled your Teleport user to access:

    apiVersion: rbac.authorization.k8s.io/v1
    kind: ClusterRoleBinding
    metadata:
      name: viewers-crb
    subjects:
    - kind: Group
      # Bind the group "viewers", corresponding to the kubernetes_groups we assigned our "kube-access" role above
      name: viewers
      apiGroup: rbac.authorization.k8s.io
    roleRef:
      kind: ClusterRole
      # "view" is a default ClusterRole that grants read-only access to resources
      # See: https://kubernetes.io/docs/reference/access-authn-authz/rbac/#user-facing-roles
      name: view
      apiGroup: rbac.authorization.k8s.io

  9. Apply the ClusterRoleBinding with kubectl:

    $ kubectl apply -f viewers-bind.yaml

Access your cluster​

When you ran the Discovery Service, it discovered your GKE cluster and registered the cluster with Teleport. You can confirm this by running the following tctl command:

$ tctl get kube_clusters
kind: kube_cluster
metadata:
  description: GKE cluster "mycluster-gke" in us-east1
  id: 0000000000000000000
  labels:
    location: us-east1
    project-id: myproject
    teleport.dev/cloud: GCP
    teleport.dev/origin: cloud
  name: mycluster-gke
spec:
  aws: {}
  azure: {}
version: v3

Run the following command to list the Kubernetes clusters that your Teleport user has access to. The list should now include your GKE cluster:

$ tsh kube ls
Kube Cluster Name   Labels                                                                                                   Selected
------------------- -------------------------------------------------------------------------------------------------------- --------
mycluster-gke location=us-east1 project-id=myproject teleport.dev/cloud=GCP teleport.dev/origin=cloud

Log in to your cluster, replacing mycluster-gke with the name of a cluster you listed previously:

$ tsh kube login mycluster-gke
Logged into kubernetes cluster "mycluster-gke". Try 'kubectl version' to test the connection.

As you can see, Teleport GKE Auto-Discovery enabled you to access a GKE cluster in your Google Cloud account without requiring you to register that cluster manually within Teleport. When you create or remove clusters in GKE, Teleport will update its state to reflect the available clusters in your account.

Troubleshooting​

Discovery Service troubleshooting​

First, check if any Kubernetes clusters have been discovered. To do this, you can use the tctl get kube_cluster command and check if the expected Kubernetes clusters have already been registered with your Teleport cluster.

If some Kubernetes clusters do not appear in the list, check if the Discovery Service selector labels match the missing Kubernetes cluster tags or look into the Discovery Service logs for permission errors.

Check that the Discovery Service is running with credentials for the correct AWS account. It can discover resources in another AWS account, but it must be configured to assume a role in the other AWS account if that's the case.

Check if there is more than one Discovery Services running:

$ tctl inventory status --connected

If you are running multiple Discovery Services, you must ensure that each service is configured with the same discovery_group value if they are watching the same cloud Kubernetes clusters or a different value if they are watching different cloud Kubernetes clusters. If this is not configured correctly, a typical symptom is kube_cluster resources being intermittently deleted from your Teleport cluster's registry.

Kubernetes Service troubleshooting​

If the tctl get kube_cluster command returns the discovered clusters, but the tctl kube ls command does not include them, check that you have set the kubernetes_service.resources section correctly.

kubernetes_service:
  enabled: "yes"
  resources:
  - labels:
      "env": "prod"

If the section is correctly configured, but clusters still do not appear or return authentication errors, please check if permissions have been correctly configured in your target cluster or that you have the correct permissions to list Kubernetes clusters in Teleport.