Skip to main content

Automatically Discover GCP Compute Instances

The Teleport Discovery Service can connect to GCP and automatically discover and enroll GCP Compute Engine instances matching configured labels. It will then execute a script on these discovered instances that will install Teleport, start it and join the cluster.

Prerequisites

  • A running Teleport cluster version 14.3.33 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 GCP compute instance to run the Discovery Service on.
  • GCP compute instances to join the Teleport cluster, running Ubuntu/Debian/RHEL if making use of the default Teleport install script. (For other Linux distributions, you can install Teleport manually.)
  • 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. tctl is supported on macOS and Linux machines. For example:
    $ tsh login --proxy=teleport.example.com --user=email@example.com
    $ tctl status
    # Cluster teleport.example.com
    # Version 14.3.33
    # 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 1/6. Create a GCP invite token

When discovering GCP compute instances, Teleport makes use of GCP invite tokens for authenticating joining SSH Service instances.

Create a file called token.yaml:

# token.yaml
kind: token
version: v2
metadata:
# the token name is not a secret because instances must prove that they are
# running in your GCP project to use this token
name: gcp-discovery-token
spec:
# use the minimal set of roles required (e.g. Node, Proxy, App, Kube, DB, WindowsDesktop)
roles: [Node]

# set the join method allowed for this token
join_method: gcp

gcp:
allow:
# The GCP project ID(s) that VMs can join from.
- project_ids: []
# (Optional) The locations that VMs can join from. Note: both regions and
# zones are accepted.
locations: []
# (Optional) The email addresses of service accounts that VMs can join
# with.
service_accounts: []

Add your instance's project ID(s) to the project_ids field. Add the token to the Teleport cluster with:

$ tctl create -f token.yaml

Step 2/6. Configure IAM permissions for Teleport

Create a service account that will give Teleport IAM permissions needed to discover instances.

Go to IAM > Roles in the GCP console and click + Create Role. Pick a name for the role (e.g. teleport-discovery) and give it the following permissions:

  • compute.instances.get
  • compute.instances.getGuestAttributes
  • compute.instances.list
  • compute.instances.setMetadata
  • iam.serviceAccounts.actAs
  • iam.serviceAccounts.get
  • iam.serviceAccounts.list

Click Create.

Go to IAM > Service accounts and click + Create Service Account. Pick a name for the service account (e.g. teleport-discovery) and copy its email address to your clipboard. Click Create and Continue.

Go to IAM and click Grant Access. Paste the service account's email into the New principals field and select your custom role. Click Save.

Step 3/6. Configure instances to be discovered

Ensure that each instance to be discovered has a service account assigned to it. No permissions are required on the service account. To check if an instance has a service account, run the following command and confirm that there is at least one entry under serviceAccounts:

$ gcloud compute instances describe --format="yaml(name,serviceAccounts)" instance_name

Enable guest attributes on instances

Guest attributes (a subset of custom metadata used for infrequently updated data) must be enabled on instances to be discovered so Teleport can access their SSH host keys. Enable guest attributes by setting enable-guest-attributes to TRUE in the instance's metadata.

$ gcloud compute instances add-metadata instance_name \
--metadata=enable-guest-attributes=True

If guest attributes are enabled during instance creation, the guest attributes will automatically be populated with the instance's host keys. If guest attributes were enabled after the instance was created, you can manually add the host keys to the guest attributes below:

Create a file named add-host-keys.sh and copy the following into it:

#!/usr/bin/env bash
for file in /etc/ssh/ssh_host_*_key.pub
do
KEY_TYPE=$(awk '{print $1}' $file)
KEY=$(awk '{print $2}' $file)
curl -X PUT --data "$KEY" "http://metadata.google.internal/computeMetadata/v1/instance/guest-attributes/hostkeys/$KEY_TYPE" -H "Metadata-Flavor: Google"
done

Run the following command to add the host keys as part of a startup script:

$ gcloud compute instances add-metadata instance_name \
--metadata-from-file=startup-script="add-host-keys.sh"

Step 4/6. Install the Teleport Discovery Service

tip

If you plan on running the Discovery Service on a host that is already running another Teleport service (Auth or Proxy, for example), you can skip this step.

Install Teleport on the virtual machine that will run the Discovery Service.

Select an edition, then follow the instructions for that edition to install Teleport.

The following command updates the repository for the package manager on the local operating system and installs the provided Teleport version:

$ curl https://cdn.teleport.dev/install-v14.3.33.sh | bash -s 14.3.33

Step 5/6. Configure Teleport to discover GCP compute instances

If you are running the Discovery Service on its own host, the service requires a valid invite token to connect to the cluster. Generate one by running the following command against your Teleport Auth Service:

$ tctl tokens add --type=discovery

Save the generated token in /tmp/token on the virtual machine that will run the Discovery Service.

In order to enable GCP instance discovery the discovery_service.gcp section of teleport.yaml must include at least one entry:

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"
gcp:
- types: ["gce"]
# The IDs of GCP projects that VMs can join from.
project_ids: []
# (Optional) The locations that VMs can join from. Note: both regions and
# zones are accepted.
locations: []
# (Optional) The email addresses of service accounts that VMs can join
# with.
service_accounts: []
# (Optional) Labels that joining VMs must have.
labels:
"env": "prod" # Match virtual machines where label:env=prod
installer:
public_proxy_addr: "teleport.example.com:443"
  • Edit the teleport.auth_server or teleport.proxy_server key to match your Auth Service or Proxy Service's domain name and port, respectively.
  • Adjust the keys under discovery_service.gcp to match your GCP environment, specifically the projects, locations, service accounts, and tags you want to associate with the Discovery Service.

GCP credentials

The Teleport Discovery Service must have the credentials of the teleport-discovery GCP service account we created above in order to be able to log in.

The easiest way to ensure that is to run the Discovery Service on a GCP instance and assign the service account to that instance. Refer to Set Up Application Default Credentials for details on alternate methods.

Step 6/6. [Optional] Customize the default installer script

To customize the default installer script, execute the following command on your workstation:

$ tctl get installer/default-installer > teleport-default-installer.yaml

The resulting teleport-default-installer.yaml can be edited to change what gets executed when enrolling discovered instances.

After making the desired changes to the default installer, the resource can be updated by executing:

$ tctl create -f teleport-default-installer.yaml

Multiple installer resources can exist and be specified in the gcp.install.script_name section of a discovery_service.gcp list item in teleport.yaml:

discovery_service:
gcp:
- types: ["gce"]
tags:
- "env": "prod"
install: # optional section when default-installer is used.
script_name: "default-installer"
- types: ["gce"]
tags:
- "env": "devel"
install:
script_name: "devel-installer"

The installer resource has the following templating options:

  • {{ .MajorVersion }}: the major version of Teleport to use when installing from the repository.
  • {{ .PublicProxyAddr }}: the public address of the Teleport Proxy Service to connect to.
  • {{ .RepoChannel }}: Optional package repository (apt/yum) channel name. Has format <channel>/<version> e.g. stable/v14. See installation for more details.
  • {{ .AutomaticUpgrades }}: indicates whether Automatic Upgrades are enabled or disabled. Its value is either true or false. See self-hosted automatic agent updates for more information.
  • {{ .TeleportPackage }}: the Teleport package to use. Its value is either teleport-ent or teleport depending on whether the cluster is enterprise or not.

These can be used as follows:

kind: installer
metadata:
name: default-installer
spec:
script: |
echo {{ .PublicProxyAddr }}
echo Teleport-{{ .MajorVersion }}
echo Repository Channel: {{ .RepoChannel }}
version: v1

Which, when retrieved for installation, will evaluate to a script with the following contents:

echo teleport.example.com
echo Teleport-14.3.33
echo Repository Channel: stable/v14.3.33

The default installer will take the following actions:

  • Add an official Teleport repository to supported Linux distributions.
  • Install Teleport via apt or yum.
  • Generate the Teleport config file and write it to /etc/teleport.yaml.
  • Enable and start the Teleport service.

Next steps