github gravitational/teleport v5.0.0
Teleport 5.0.0
on GitHub

latest releases: v16.0.0-dev.fred-manifests.7, api/v16.0.0-dev.fred-manifests.7, v16.0.0-dev.fred-manifests.6...
3 years ago

Teleport 5.0 is a major release with new features, functionality, and bug fixes. Users can review 5.0 closed issues on Github for details of all items.

New Features

Teleport 5.0 introduces two distinct features: Teleport Application Access and significant Kubernetes Access improvements - multi-cluster support.

Teleport Application Access

Teleport can now be used to provide secure access to web applications. This new feature was built with the express intention of securing internal apps which might have once lived on a VPN or had a simple authorization and authentication mechanism with little to no audit trail. Application Access works with everything from dashboards to single page Javascript applications (SPA).

Application Access uses mutually authenticated reverse tunnels to establish a secure connection with the Teleport unified Access Plane which can then becomes the single ingress point for all traffic to an internal application.

Adding an application follows the same UX as adding SSH servers or Kubernetes clusters, starting with creating a static or dynamic invite token. 

$ tctl tokens add --type=app

Then simply start Teleport with a few new flags. 

$ teleport start --roles=app --token=xyz --auth-server=proxy.example.com:3080 \
    --app-name="example-app" \
    --app-uri="http://localhost:8080"

This command will start an app server that proxies the application "example-app" running at http://localhost:8080 at the public address https://example-app.example.com.

Applications can also be configured using the new app_service section in teleport.yaml. 

app_service:
   # Teleport Application Access is enabled.
   enabled: yes
   # We've added a default sample app that will check
   # that Teleport Application Access is working
   # and output JWT tokens.
   # https://dumper.teleport.example.com:3080/
   debug_app: true
   apps:
   # Application Access can be used to proxy any HTTP endpoint.
   # Note: Name can't include any spaces and should be DNS-compatible A-Za-z0-9-._
   - name: "internal-dashboard"
     uri: "http://10.0.1.27:8000"
     # By default Teleport will make this application
     # available on a sub-domain of your Teleport proxy's hostname
     # internal-dashboard.teleport.example.com
     # - thus the importance of setting up wilcard DNS.
     # If you want, it's possible to set up a custom public url.
     # DNS records should point to the proxy server.
     # internal-dashboard.teleport.example.com
     # Example Public URL for the internal-dashboard app.
     # public_addr: "internal-dashboard.acme.com"
     # Optional labels
     # Labels can be combined with RBAC rules to provide access.
     labels:
       customer: "acme"
       env: "production"
     # Optional dynamic labels
     commands:
     - name: "os"
       command: ["/usr/bin/uname"]
       period: "5s"
     # A proxy can support multiple applications. Application Access
     # can also be deployed with a Teleport node.
     - name: "arris"
       uri: "http://localhost:3001"
       public_addr: "arris.example.com"

Application access requires two additional changes. DNS must be updated to point the application domain to the proxy and the proxy must be loaded with a TLS certificate for the domain. Wildcard DNS and TLS certificates can be used to simplify deployment. 

# When adding the app_service certificates are required to provide a TLS
# connection. The certificates are managed by the proxy_service
proxy_service:
  # We've extended support for https certs. Teleport can now load multiple
  # TLS certificates. In the below example we've obtained a wildcard cert
  # that'll be used for proxying the applications.
  # The correct certificate is selected based on the hostname in the HTTPS
  # request using SNI.
  https_keypairs:
  - key_file: /etc/letsencrypt/live/teleport.example.com/privkey.pem
    cert_file: /etc/letsencrypt/live/teleport.example.com/fullchain.pem
  - key_file: /etc/letsencrypt/live/*.teleport.example.com/privkey.pem
    cert_file: /etc/letsencrypt/live/*.teleport.example.com/fullchain.pem

You can learn more at https://goteleport.com/teleport/docs/application-access/

Teleport Kubernetes Access

Teleport 5.0 also introduces two highly requested features for Kubernetes.

  • The ability to connect multiple Kubernetes Clusters to the Teleport Access Plane, greatly reducing operational complexity.
  • Complete Kubernetes audit log capture #4526, going beyond the existing kubectl exec capture.

For a full overview please review the Kubernetes RFD.

To support these changes, we've introduced a new service. This moves Teleport Kubernetes configuration from the proxy_service into its own dedicated kubernetes_service section.

When adding the new Kubernetes service, a new type of join token is required. 

tctl tokens add --type=kube

Example configuration for the new kubernetes_service: 

# ...
kubernetes_service:
   enabled: yes
   listen_addr: 0.0.0.0:3027
   kubeconfig_file: /secrets/kubeconfig

Note: a Kubernetes port still needs to be configured in the proxy_service via kube_listen_addr.

New "tsh kube" commands

tsh kube commands are used to query registered clusters and switch kubeconfig context: 

$ tsh login --proxy=proxy.example.com --user=awly

# list all registered clusters
$ tsh kube ls
Cluster Name       Status
-------------      ------
a.k8s.example.com  online
b.k8s.example.com  online
c.k8s.example.com  online

# on login, kubeconfig is pointed at the first cluster (alphabetically)
$ kubectl config current-context
proxy.example.com-a.k8s.example.com

# but all clusters are populated as contexts
$ kubectl config get-contexts
CURRENT   NAME                     CLUSTER             AUTHINFO
*         proxy.example.com-a.k8s.example.com   proxy.example.com   proxy.example.com-a.k8s.example.com
          proxy.example.com-b.k8s.example.com   proxy.example.com   proxy.example.com-b.k8s.example.com
          proxy.example.com-c.k8s.example.com   proxy.example.com   proxy.example.com-c.k8s.example.com

# switch between different clusters:
$ tsh kube login c.k8s.example.com

# the traditional way is also supported:
$ kubectl config use-context proxy.example.com-c.k8s.example.com

# check current cluster
$ kubectl config current-context
proxy.example.com-c.k8s.example.com

Other Kubernetes changes:

  • Support k8s clusters behind firewall/NAT using a single Teleport cluster #3667
  • Support multiple k8s clusters with a single Teleport proxy instance #3952

Additional User and Token Resource

We've added two new RBAC resources; these provide the ability to limit token creation and to list and modify Teleport users: 

- resources: [user]
  verbs: [list,create,read,update,delete]
- resources: [token]
  verbs: [list,create,read,update,delete]

Learn more about Teleport's RBAC Resources

Cluster Labels

Teleport 5.0 also adds the ability to set labels on Trusted Clusters. The labels are set when creating a trusted cluster invite token. This lets teams use the same RBAC controls used on nodes to approve or deny access to clusters. This can be especially useful for MSPs that connect hundreds of customers' clusters - when combined with Access Workflows, cluster access can easily be delegated. Learn more by reviewing our Truster Cluster Setup & RBAC Docs

Creating a trusted cluster join token for a production environment: 

$ tctl tokens add --type=trusted_cluster --labels=env=prod
kind: role
#...
  deny:
    # cluster labels control what clusters user can connect to. The wildcard ('*')
    # means any cluster. By default, deny rules are empty to preserve backwards
    # compatibility
    cluster_labels:
      'env': 'prod'

Teleport UI Updates

Teleport 5.0 also iterates on the UI Refresh from 4.3. We've moved the cluster list into our sidebar and have added an Application launcher. For customers moving from 4.4 to 5.0, you'll notice that we have moved session recordings back to their own dedicated section.

Other updates:

  • We now provide local user management via https://[cluster-url]/web/users, providing the ability to easily edit, reset and delete local users.
  • Teleport Node & App Install scripts. This is currently an Enterprise-only feature that provides customers with an easy 'auto-magic' installer script. Enterprise customers can enable this feature by modifying the 'token' resource. See note above.
  • We've added a Waiting Room for customers using Access Workflows. Docs

Signed RPM and Releases

Starting with Teleport 5.0, we now provide an RPM repo for stable releases of Teleport. We've also started signing our RPMs to provide assurance that you're always using an official build of Teleport.

See https://rpm.releases.teleport.dev/ for more details.

Improvements

  • Added --format=json playback option for tsh play. For example tsh play --format=json ~/play/0c0b81ed-91a9-4a2a-8d7c-7495891a6ca0.tar | jq '.event can be used to show all events within an a local archive. #4578
  • Added support for continuous backups and auto scaling for DynamoDB. #4780
  • Added a Linux ARM64/ARMv8 (64-bit) Release. #3383
  • Added https_keypairs field which replaces https_key_file and https_cert_file. This allows administrators to load multiple HTTPS certs for Teleport Application Access. Teleport 5.0 is backwards compatible with the old format, but we recommend updating your configuration to use https_keypairs.

Enterprise Only:

  • tctl can load credentials from ~/.tsh #4678
  • Teams can require a user submitted reason when using Access Workflows #4573

Fixes

  • Updated tctl to always format resources as lists in JSON/YAML. #4281
  • Updated tsh status to now print Kubernetes status. #4348
  • Fixed intermittent issues with loginuid.so. #3245
  • Reduced access denied to Proxy log spam. #2920
  • Various AMI fixes: paths are now consistent with other Teleport packages and configuration files will not be overwritten on reboot.

Documentation

We've added an API Reference to simply developing applications against Teleport.

Upgrade Notes

Please follow our standard upgrade procedure.

  • Optional: Consider updating https_key_file & https_cert_file to our new https_keypairs: format.
  • Optional: Consider migrating Kubernetes Access from proxy_service to kubernetes_service after the upgrade.
Check out latest releases or
releases around gravitational/teleport v5.0.0

Don't miss a new teleport release

NewReleases is sending notifications on new releases.

Get notifications