Giter VIP home page Giter VIP logo

kube-oidc-proxy's Introduction

kube-oidc-proxy

kube-oidc-proxy is a reverse proxy server to authenticate users using OIDC to Kubernetes API servers where OIDC authentication is not available (i.e. managed Kubernetes providers such as GKE, EKS, etc).

This intermediary server takes kubectl requests, authenticates the request using the configured OIDC Kubernetes authenticator, then attaches impersonation headers based on the OIDC response from the configured provider. This impersonated request is then sent to the API server on behalf of the user and it's response passed back. The server has flag parity with secure serving and OIDC authentication that are available with the Kubernetes API server as well as client flags provided by kubectl. In-cluster client authentication is also available when running kube-oidc-proxy as a pod.

Since the proxy server utilises impersonation to forward requests to the API server once authenticated, impersonation is disabled for user requests to the API server.

kube-oidc-proxy demo

The following is a diagram of the request flow for a user request. kube-oidc-proxy request flow

Quickest Start

OpenUnison integrates kube-oidc-proxy directly, and includes an identity provider and access portal for Kubernetes. The quickest way to get started with kube-oidc-proxy is to follow the directions for OpenUnison's deployment at https://openunison.github.io/.

Tutorial

Directions on how to deploy OIDC authentication with multi-cluster can be found here. or there is a helm chart.

Quickstart

Deployment yamls can be found in ./deploy/yaml and will require configuration to an exiting OIDC issuer.

This quickstart demo will assume you have a Kubernetes cluster without OIDC authentication, as well as an OIDC client created with your chosen provider. We will be using a Service with type LoadBalancer to expose it to the outside world. This can be changed depending on what is available and what suites your set up best.

Firstly deploy kube-oidc-proxy and it's related resources into your cluster. This will create it's Deployment, Service Account and required permissions into the newly created kube-oidc-proxy Namespace.

$ kubectl apply -f ./deploy/yaml/kube-oidc-proxy.yaml
$ kubectl get all --namespace kube-oidc-proxy

This deployment will fail until we create the required secrets. Notice we have also not provided any client flags as we are using the in-cluster config with it's Service Account.

We now wait until we have an external IP address provisioned.

$ kubectl get service --namespace kube-oidc-proxy

We need to generate certificates for kube-oidc-proxy to securely serve. These certificates can be generated through cert-manager, more information about this project found here.

Next, populate the OIDC authenticator Secret using the secrets given to you by your OIDC provider in ./deploy/yaml/secrets.yaml. The OIDC provider CA will be different depending on which provider you are using. The easiest way to obtain the correct certificate bundle is often by opening the providers URL into a browser and fetching them there (typically output by clicking the lock icon on your address bar). Google's OIDC provider for example requires CAs from both https://accounts.google.com/.well-known/openid-configuration and https://www.googleapis.com/oauth2/v3/certs.

Apply the secret manifests.

kubectl apply -f ./deploy/yaml/secrets.yaml

You can restart the kube-oidc-proxy pod to use these new secrets now they are available.

kubectl delete pod --namespace kube-oidc-proxy kube-oidc-proxy-*

Finally, create a Kubeconfig to point to kube-oidc-proxy and set up your OIDC authenticated Kubernetes user.

apiVersion: v1
clusters:
- cluster:
    certificate-authority: *
    server: https://[url|ip:443]
  name: *
contexts:
- context:
    cluster: *
    user: *
  name: *
kind: Config
preferences: {}
users:
- name: *
  user:
    auth-provider:
      config:
        client-id: *
        client-secret: *
        id-token: *
        idp-issuer-url: *
        refresh-token: *
      name: oidc

Configuration

Logging

In addition to auditing, kube-oidc-proxy logs all requests to standard out so the requests can be captured by a common Security Information and Event Management (SIEM) system. SIEMs will typically import logs directly from containers via tools like fluentd. This logging is also useful in debugging. An example successful event:

[2021-11-25T01:05:17+0000] AuSuccess src:[10.42.0.5 / 10.42.1.3, 10.42.0.5] URI:/api/v1/namespaces/openunison/pods?limit=500 inbound:[mlbadmin1 / system:masters|system:authenticated /]

The first block, between [] is an ISO-8601 timestamp. The next text, AuSuccess, indicates that authentication was successful. the src block containers the remote address of the request, followed by the value of the X-Forwarded-For HTTP header if provided. The URI is the URL path of the request. The inbound section provides the user name, groups, and extra-info provided to the proxy from the JWT.

When there's an error or failure:

[2021-11-25T01:05:24+0000] AuFail src:[10.42.0.5 / 10.42.1.3] URI:/api/v1/nodes

This is similar to success, but without the token information.

End-User Impersonation

kube-oidc-proxy supports the impersonation headers for inbound requests. This allowes the proxy to support kubectl --as. When impersonation headers are included in a request, the proxy checks that the authenticated user is able to assume the identity of the impersonation headers by submitting SubjectAccessReview requests to the API server. Once authorized, the proxy will send those identity headers instead of headers generated for the authenticated user. In addition, three Extra impersonation headers are sent to the API server to identify the authenticated user who's making the request:

Header Description
originaluser.jetstack.io-user The original username
originaluser.jetstack.io-groups The original groups
originaluser.jetstack.io-extra A JSON encoded map of arrays representing all of the extra headers included in the original identity

In addition to sending this extra information, the proxy adds an additional section to the logfile that will identify outbound identity data. When impersonation headers are present, the AuSuccess log will look like:

[2021-11-25T01:05:17+0000] AuSuccess src:[10.42.0.5 / 10.42.1.3] URI:/api/v1/namespaces/openunison/pods?limit=500 inbound:[mlbadmin1 / system:masters|system:authenticated /] outbound:[mlbadmin2 / group2|system:authenticated /]

When using Impersonate-Extra- headers, the proxy's ServiceAccount must be explicitly authorized via RBAC to impersonate whatever the extra key is named. This is because extras are treated as subresources which must be explicitly authorized.

Development

NOTE: building kube-oidc-proxy requires Go version 1.17 or higher.

To help with development, there is a suite of tools you can use to deploy a functioning proxy from source locally. You can read more here.

kube-oidc-proxy's People

Contributors

joshvanl avatar mlbiam avatar jetstack-bot avatar simonswine avatar mhrabovcin avatar mattbates avatar dcherman avatar sellitforcache avatar phanama avatar allanhung avatar x-way avatar justinas-b avatar ltagliamonte-dd avatar wallrj avatar alejandroesc avatar binboum avatar saiteja313 avatar

Watchers

James Cloos avatar  avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.