architecture-center/doc/blueprints/source/best-practices/security/deploy_keycloak.rst

Deploy Keycloak on CCE

Overview

Keycloak is an open-source identity and access management (IAM) solution developed by Red Hat. It provides features for single sign-on (SSO), user authentication, authorization, and identity brokering. Keycloak aims to simplify the implementation of authentication and authorization mechanisms in applications by offering a centralized and configurable platform.

Key features of Keycloak include:

1. Single Sign-On (SSO): Keycloak enables users to log in once and gain access to multiple applications without the need to re-enter credentials for each application.
2. Identity Federation: It supports identity brokering, allowing users to log in with existing accounts from social networks (such as Google, Facebook, or GitHub) or other identity providers.
3. User Authentication: Keycloak provides a variety of authentication mechanisms, including username and password, multi-factor authentication, and support for external identity providers.
4. Authorization Services: It includes fine-grained access control and authorization policies to manage what users can and cannot do within applications.
5. User Account Management: Keycloak offers user self-registration, password reset, and other account management features.
6. LDAP and Active Directory Integration: It supports integration with LDAP (Lightweight Directory Access Protocol) and Microsoft Active Directory for seamless user management.
7. Client Adapters: Keycloak provides client adapters for various platforms and languages, making it easier to integrate with applications built using different technologies.
8. Security and Compliance: Keycloak follows best practices for security and compliance, including support for OAuth 2.0 and OpenID Connect standards.

Developers can integrate Keycloak with their applications using various protocols such as OpenID Connect, OAuth 2.0, SAML (Security Assertion Markup Language), and more. It is commonly used in microservices architectures and distributed systems to manage authentication and authorization in a centralized manner. Keycloak is often employed in scenarios where secure user authentication and access control are crucial, such as enterprise applications, web applications, and APIs.

In this blueprint, we are going to discuss the steps to install Keycloak, in Open Telekom Cloud, on a CCE Cluster.

Create a VPC and a Subnet

We are going to need a Virtual Private Cloud (VPC) and at least one Subnet where we are going to provision both RDS instances and CCE nodes. For enhanced security granularity, we could split those resources in two different Subnets.

Warning

RDS and CCE nodes have to be on the same VPC.

Deploy a PostgreSQL with RDS

Keycloak, as a stateful workload, requires the presence of a persistent storage in order to maintain its data and configuration during pod restarts. We could deploy a PostgreSQL database as a CCE workload, but this would require additional administrative overhead from your side. The Managed Relational Database Service of Open Telekom Cloud is a perfect fit for this scenario. A scalable turn-key solution, that fully integrated with the rest of managed services of the platform without demanding from the consumer additional administrative effort.

Create Security Groups

We are going to need two different Security Groups. One for the RDS nodes, so it can accept client calls on port 5432 (Inbound Rules), which they only reside in the same Subnet (in case we went for a single Subnet solution):

And one Security Group for the client nodes that need to access the database (Outbound Rules), in our case those would be the CCE nodes where Keycloak is going to be installed on.

Provision a Database

Now as next, we need to provision a PostgreSQL 14 database. Pick the instance and storage class size that fit your needs:

and make sure that you:

• you place the RDS nodes in the same VPC with the CCE nodes
• assign rds-instances as the Security Group for the RDS nodes

Create a Private DNS Zone

We are provisioning PostgreSQL in order to support the functionality of Keycloak. For that matter, although Open Telekom Cloud employs this RDS instance with a floating IP address, it would be better that we connect the RDS instance with Keycloak via a fully qualified domain name and let the Open Telekom Cloud's DNS service to manage the resolution of that endpoints. In the Domain Name Service management panel click Private Zone and create a new one that points to the VPC that CCE and RDS nodes are placed:

and then click Manage Record Set to add a new A Record to this zone:

Note

The domain name, will be a fictitious domain representing your solution and not a public one. It can be virtually any domain or subdomain that conforms to the a FQDN rules.

The floating IP of the RDS instance can be found in the Basic Information panel of the database:

Provision a CCE Cluster

We are going to need a CCE Cluster. In order to provision one, you can follow the configuration steps of the wizard paying attention to the following details:

• We are not going to need an HA cluster - of course adjust to your needs because this is not something you can change in the future.
• We need to provision the CCE Cluster in the same VPC as the RDS nodes.
• If you follow the single Subnet lab instructions make sure you place the CCE Nodes in the same Subnet that RDS nodes reside.

Add worker nodes to the CCE cluster using the wizard, and wait all nodes to become operational. Then add to each node an additional Security Group, in particular the rds-client that we created earlier in this lab.

Note

Make your own decision how you're going to access this CCE Cluster afterwards. You can assign an Elastic IP Address and access it over the Internet or provision and additional public-facing bastion host and access it through this machine. We categorically recommend the latter.

Deploy Keycloak

We are going to deploy Keycloak using simple Kubernetes manifests. Deploy those YAML manifests in the order described below using the command on your bastion host (or in any other machine if you chose to go for an EIP):

kubectl apply -f <<filename.yaml>>

Deploy Keycloak Secrets

First we are going to need a Namespace in our CCE Cluster, in order to deploy all the resources required by Keycloak:

We are going to need two Secrets. One, postgres-credentials, that will contain the credentials to access the PostgreSQL database instance and a second one, keycloak-secrets, that will contain the necessary credential to access the web console of Keycloak:

apiVersion: v1
kind: Secret
metadata:
  name: postgres-credentials
  namespace: keycloak
type: Opaque
stringData:
  POSTGRES_USER: root
  POSTGRES_PASSWORD: <<POSTGRES_PASSWORD>>
  POSTGRES_DB: postgres
---
apiVersion: v1
kind: Secret
metadata:
  name: keycloak-secrets
  namespace: keycloak
type: Opaque
stringData:
  KEYCLOAK_ADMIN: admin
  KEYCLOAK_ADMIN_PASSWORD: <<KEYCLOAK_ADMIN_PASSWORD>>

Note

POSTGRES_PASSWORD is the password for the root user your provided during the creation of the RDS instance.

KEYCLOAK_ADMIN_PASSWORD, as we mentioned before, is the password for the admin user of the Keycloak web console. You can easily create random strong passwords, in Linux terminal, with the following command:

openssl rand -base64 14

Deploy Keycloak Application

Next step, is deploying Keycloak itself:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: keycloak
  namespace: keycloak
  labels:
    app: keycloak
spec:
  replicas: 1
  selector:
    matchLabels:
      app: keycloak
  template:
    metadata:
      labels:
        app: keycloak
    spec:
      containers:
      - name: keycloak
        image: quay.io/keycloak/keycloak:21.0.2
        args: ["start-dev"]
        env:
        - name: KEYCLOAK_ADMIN
          valueFrom:
            secretKeyRef:
              key: KEYCLOAK_ADMIN
              name: keycloak-secrets
        - name: KEYCLOAK_ADMIN_PASSWORD
          valueFrom:
            secretKeyRef:
              key: KEYCLOAK_ADMIN_PASSWORD
              name: keycloak-secrets
        - name: KC_PROXY
          value: "edge"
        - name: KC_HEALTH_ENABLED
          value: "true"
        - name: KC_METRICS_ENABLED
          value: "true"
        - name: KC_HOSTNAME_STRICT_HTTPS
          value: "true"
        - name: KC_LOG_LEVEL
          value: INFO
        - name: KC_DB
          value: postgres
        - name: POSTGRES_DB
          valueFrom:
            secretKeyRef:
              name: postgres-credentials
              key: POSTGRES_DB
        - name: KC_DB_URL
          value: jdbc:postgresql://postgresql.blueprints.arc:5432/$(POSTGRES_DB)
        - name: KC_DB_USERNAME
          valueFrom:
            secretKeyRef:
              name: postgres-credentials
              key: POSTGRES_USER
        - name: KC_DB_PASSWORD
          valueFrom:
            secretKeyRef:
              name: postgres-credentials
              key: POSTGRES_PASSWORD
        ports:
        - name: http
          containerPort: 8080
        readinessProbe:
          httpGet:
            path: /health/ready
            port: 8080
          initialDelaySeconds: 250
          periodSeconds: 10
        livenessProbe:
          httpGet:
            path: /health/live
            port: 8080
          initialDelaySeconds: 500
          periodSeconds: 30
        resources:
            limits:
              memory: 512Mi
              cpu: "1"
            requests:
              memory: 256Mi
              cpu: "0.2"

As you will notice in the highlighted lines, we parameterize the credentials portion of this manifest by referencing the variables and their values we installed in the previous step with the Secrets. Important to mention the significance of line 51, where we connect Keycloak with the RDS instance using the FQDN we created in our Private DNS Zone for this instance.

Deploy Keycloak Service

We deployed the application, but at the time being is not accessible by an internal or external actor (direct access from Pods does not count in this case). For that matter, we need to deploy a Service that will expose Keycloak's workload:

apiVersion: v1
kind: Service
metadata:
  name: keycloak
  namespace: keycloak
  labels:
    app: keycloak
spec:
  ports:
  - name: https
    port: 443
    targetPort: 8080
  selector:
    app: keycloak
  type: NodePort

Note

Pay attention to line 15, where we set the type as NodePort. That's because we want to expose this service externally, in a later step, via an Ingress.

Expose Keycloak

Create an Elastic Load Balancer

First in our list for this part, is to create an Elastic Load Balancer that will be employed with the following:

• An EIP address
• Support L4 and L7 load balancing
• Be in the same VPC/Subnet as the nodes of our CCE Cluster
• Associate backend servers by using their IP addresses (IP as Backend)

Note

Note down the ELB ID, we are going to need it to configure the Nginx Ingress that we will deploy next.

Deploy Nginx Ingress on CCE

We are going to deploy in this step the Ingress that will sit between our ELB and the Keycloak Service and expose it in the address of our preference (keycloak.example.com for this lab)

Warning

Do not forget that the FQDN we are going to use to expose the Keycloak Service has to point to a real domain or subdomain that you actually own!

We will use Helm to deploy Nginx Ingress to our CCE Cluster. Helm is the de-facto package manager of Kubernetes and if you don't have it already installed on your remote machine or your bastion host, you can do it with the following commands:

curl -fsSL -o get_helm.sh https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3
chmod 700 get_helm.sh
./get_helm.sh

We have to provide to the helm chart a couple configuration values (overrides.yaml), among them the internal ID of the Elastic Load Balancer is the most important - as it will bind the future ingresses that will be created using this ingress class with the specific load balancer.

controller:
  replicaCount: 1
  service:
    externalTrafficPolicy: Cluster
    annotations:
      kubernetes.io/elb.id: "0000000-0000-0000-0000-000000000000"

Note

Special attention required at line 6, replace the placeholder value with the ID you copied from the main panel of your newly created Elastic Load Balancer.

We can now install the chart (it will automatically create and deploy everything in a namespace named nginx-system):

helm upgrade --install -f overrides.yaml --install ingress-nginx ingress-nginx \
--repo https://kubernetes.github.io/ingress-nginx \
--namespace nginx-system --create-namespace

Create a Public DNS Endpoint

As we will see later, when we will reach to the point that we are ready to register this Keycloak installation as an Identity Provider (IdP) in our Open Telekom Cloud tenant, it is really pertinent that the EIP of our ELB resolves to a real, secure URL address:

In order to accomplish that, we have to transfer the management of the NS-Records of your domain to the Domain Name Service of Open Telekom Cloud. Go on the site of your registar and make sure you configure the following:

• Turn off any dynamic dns service for the domain or the subdomains you are going to bind with Keycloak.
• Change the NS-Records of your domain to point to: ns1.open-telekom-cloud.com and ns2.open-telekom-cloud.com

If those two prerequisites are met, then you are ready to configure a new DNS Public Zone and Record Sets for your domain in Open Telekom Cloud. We do have two mutually exclusive options to do that:

• Create manually from Open Telekom Cloud Console, a new Public DNS Zone that binds to your domain and an A-Record in that zone that points to the EIP of the external load balancer.
• Automate everything using ExternalDNS.

Create the Endpoint manually

Follow the same steps we did earlier for the Private Zone, but this time create a Public Zone targeting to your domain and add an A-Record that binds your Keycloak's (sub)domain with the Elastic IP Address of the Elastic Load Balancer.

Create the Endpoint with ExternalDNS

What is ExternalDNS? Quoting directly from the official repo of the project:

Inspired by Kubernetes DNS, Kubernetes' cluster-internal DNS server, ExternalDNS makes Kubernetes resources discoverable via public DNS servers. Like KubeDNS, it retrieves a list of resources (Services, Ingresses, etc.) from the Kubernetes API to determine a desired list of DNS records. Unlike KubeDNS, however, it's not a DNS server itself, but merely configures other DNS providers accordingly—e.g. AWS Route 53 or Google Cloud DNS.

In a broader sense, ExternalDNS allows you to control DNS records dynamically via Kubernetes resources in a DNS provider-agnostic way.

Deploy ExternalDNS on CCE

We are going to deploy ExternalDNS with Helm as well. First let's lay down the configuration of the chart in a file name overrides.yaml:

sources:
  - crd
  - service
  - ingress
provider: designate
combineFQDNAnnotation: true
crd:
  create: true
logFormat: json
designate:
  username: "OTCAC_DNS_ServiceAccount"
  password: <<OTCAC_DNS_ServiceAccount_PASSWORD>>
  authUrl: "https://iam.eu-de.otc.t-systems.com:443/v3"
  regionName: "eu-de"
  userDomainName: "OTCXXXXXXXXXXXXXXXXXXXX"
  projectName: "eu-de_XXXXXXXXXXX"

Warning

Special attention required at lines 13,14. Although DNS is a global service, all changes have to be applied in Region eu-de.

Install the chart (it will deploy all the necessary resources in an automatically created namespace called external-dns:

helm repo add bitnami https://charts.bitnami.com/bitnami
helm repo update

helm upgrade --install -f overrides.yaml external-dns bitnami/external-dns -n external-dns --create-namespace

Create a dedicated DNS Service Account

Note

This is required only when ExternalDNS is used.

Go to IAM management console, and create a new User that permits programmatic access to Open Telekom Cloud resources:

Grant this User the following permissions or add him directly to User Group dns-admins (if it exists)

Deploy a Keycloak Endpoint

We have now laid all the groundwork in order to automatically provision a Public DNS Zone and a dedicated A-Record that will bind the EIP of our ELB with Keycloak's subdomain FQDN. For that matter we need to install a Custom Resource based on a CRD installed by ExternalDNS that is called DNSEndpoint:

apiVersion: externaldns.k8s.io/v1alpha1
kind: DNSEndpoint
metadata:
  name: keycloak
  namespace: keycloak
spec:
  endpoints:
  - dnsName: keycloak.example.de
    recordTTL: 300
    recordType: A
    targets:
    - XXX.XXX.XXX.XXX

Note

At line 12, replace the placeholder with the Elastic IP Address that is assigned to your Elastic Load Balancer. At line 8, replace the (sub)domain with the one of yours

Wait for a couple of seconds, till the reconciliation loop of the ExternalDNS controller is done, and if all went well you should now see the Record Sets of your Public Zone populated with various entries:

Deploy Keycloak Ingress

And finally, the last step of this lab is to deploy an ingress for the Keycloak Service:

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: keycloak-ingress
  namespace: keycloak
  annotations:
    nginx.ingress.kubernetes.io/rewrite-target: /
    kubernetes.io/ingress.class: nginx
spec:
  rules:
  - http:
      paths:
      - path: /
        pathType: Prefix
        backend:
          service:
            name: keycloak
            port:
              number: 443

We can now open the url address we defined in our Public DNS Zone for this application and land on the welcome page of Keycloak:

Next Steps

• Identity Federation through Keycloak/GitHub