Consul Service Discovery and Mesh on Minikube

In this tutorial, you'll start a local Kubernetes cluster with Minikube. You will then deploy Consul with the official Helm chart. After deploying Consul, you will learn how to access the agents. You will then deploy two services that use Consul to discover each other.

»Prerequisites

First, you'll need to follow the directions for installing Minikube, including VirtualBox or a similar virtualization tool.

You'll also need to install kubectl and helm.

$ brew install kubernetes-cli

$ brew install kubernetes-cli

$ brew install kubernetes-helm

$ brew install kubernetes-helm

»Start Minikube

Start Minikube with the optional --memory flag specifying the equivalent of 4-8GB of memory, so your pods will have plenty of resources to use. Starting Minikube may take several minutes. It will download a 100-300MB of dependencies and container images.

$ minikube start --memory 4096

$ minikube start --memory 4096

»Install Consul with the official Helm chart

»Download the demo code and Helm chart

First, add the HashiCorp Helm Chart repository:

$ helm repo add hashicorp https://helm.releases.hashicorp.com
"hashicorp" has been added to your repositories

$ helm repo add hashicorp https://helm.releases.hashicorp.com"hashicorp" has been added to your repositories

»Create a custom values file

The chart comes with reasonable defaults, however, you will override a few values to help things go more smoothly with kind and enable useful features.

Create a custom values file called config.yaml with the following contents. This configuration will:

• Set the prefix used for all resources in the Helm chart to consul
• Name the Consul datacenter dc1
• Configure the datacenter to run only 1 server
• Configure the server to use the root user
• Enable the Consul UI and expose it via a NodePort
• Enable Consul service mesh features by setting connectInject.enabled to true
• Enable Consul service mesh CRDs by setting controller.enabled to true

$ cat > config.yaml <<EOF
global:
  name: consul
  datacenter: dc1
server:
  replicas: 1
  securityContext:
    runAsNonRoot: false
    runAsGroup: 0
    runAsUser: 0
    fsGroup: 0
ui:
  enabled: true
  service:
    type: 'NodePort'
connectInject:
  enabled: true
controller:
  enabled: true
EOF

$ cat > config.yaml <<EOFglobal:  name: consul  datacenter: dc1server:  replicas: 1  securityContext:    runAsNonRoot: false    runAsGroup: 0    runAsUser: 0    fsGroup: 0ui:  enabled: true  service:    type: 'NodePort'connectInject:  enabled: truecontroller:  enabled: trueEOF

»Deploy Consul on Minikube

Now, run helm install, providing your custom values file, the hashicorp/consul chart, and a name for your Consul installation. It will print a list of all the resources that were created.

$ helm install -f config.yaml consul hashicorp/consul
NAME: hashicorp
...TRUNCATED...
To learn more about the release if you are using Helm 3, run:

  $ helm status hashicorp
  $ helm get all hashicorp

$ helm install -f config.yaml consul hashicorp/consulNAME: hashicorp...TRUNCATED...To learn more about the release if you are using Helm 3, run:
  $ helm status hashicorp  $ helm get all hashicorp

»Access the Consul UI

Verify Consul was deployed properly by accessing the Consul UI. Run minikube service list to list your services. Find the one with consul-ui in the name.

$ minikube service list
|-------------|-----------------------------|--------------|----------------------------|
|  NAMESPACE  |            NAME             | TARGET PORT  |            URL             |
|-------------|-----------------------------|--------------|----------------------------|
| default     | consul-connect-injector-svc | No node port |
| default     | consul-controller-webhook   | No node port |
| default     | consul-dns                  | No node port |
| default     | consul-server               | No node port |
| default     | consul-ui                   | http/80      | http://192.168.64.32:30881 |
| default     | kubernetes                  | No node port |
| kube-system | kube-dns                    | No node port |
|-------------|-----------------------------|--------------|----------------------------|

$ minikube service list|-------------|-----------------------------|--------------|----------------------------||  NAMESPACE  |            NAME             | TARGET PORT  |            URL             ||-------------|-----------------------------|--------------|----------------------------|| default     | consul-connect-injector-svc | No node port || default     | consul-controller-webhook   | No node port || default     | consul-dns                  | No node port || default     | consul-server               | No node port || default     | consul-ui                   | http/80      | http://192.168.64.32:30881 || default     | kubernetes                  | No node port || kube-system | kube-dns                    | No node port ||-------------|-----------------------------|--------------|----------------------------|

Run minikube service with the consul-ui service name as the argument. It will open the service in your web browser.

$ minikube service consul-ui

$ minikube service consul-ui

You can now visit the Consul UI with a list of Consul's services, nodes, and other resources. Currently, you should only find the consul service listed.

»Access Consul with kubectl and the HTTP API

In addition to accessing Consul with the UI, you can manage Consul with the HTTP API or by directly connecting to the pod with kubectl.

»Use Kubectl to access the server

To access the Consul server pod you can start a shell session in the pod with kubectl exec.

$ kubectl exec -it consul-server-0 -- /bin/sh

$ kubectl exec -it consul-server-0 -- /bin/sh

Your terminal prompt will change to the following, which indicates you have successfully started an interactive terminal session in the server container.

/ #

/ #

From the container, you are able to navigate the file system and run Consul CLI commands. For example, you can view the Consul version.

$ consul version
Consul v1.9.4
Revision 10bb6cb3b
Protocol 2 spoken by default, understands 2 to 3 (agent will automatically use protocol >2 when speaking to compatible agents)

$ consul versionConsul v1.9.4Revision 10bb6cb3bProtocol 2 spoken by default, understands 2 to 3 (agent will automatically use protocol >2 when speaking to compatible agents)

You can also view members of the datacenter.

$ consul members
Node             Address          Status  Type    Build  Protocol  DC   Segment
consul-server-0  172.17.0.6:8301  alive   server  1.9.4  2         dc1  <all>
minikube         172.17.0.5:8301  alive   client  1.9.4  2         dc1  <default>

$ consul membersNode             Address          Status  Type    Build  Protocol  DC   Segmentconsul-server-0  172.17.0.6:8301  alive   server  1.9.4  2         dc1  <all>minikube         172.17.0.5:8301  alive   client  1.9.4  2         dc1  <default>

Type exit and then enter to leave the container session.

$ exit

$ exit

»Consul HTTP API

You can use the Consul HTTP API by communicating with the local agent running on the Kubernetes node. Read the documentation to learn more about using the Consul HTTP API with Kubernetes.

»Deploy services with Kubernetes

Now that you have a running Consul service mesh, you can deploy services to it.

»Deploy two services

You will now deploy a two-tier application made of a backend data service that returns a number (the counting service), and a frontend dashboard that pulls from the counting service over HTTP and displays the number.

Create a deployment definition, service, and service account for the counting service named counting.yaml.

$ cat > counting.yaml <<EOF
apiVersion: v1
kind: ServiceAccount
metadata:
  name: counting
---
apiVersion: v1
kind: Service
metadata:
  name: counting
spec:
  selector:
    app: counting
  ports:
  - port: 9001
    targetPort: 9001
---
apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: counting
  name: counting
spec:
  replicas: 1
  selector:
    matchLabels:
      app: counting
  template:
    metadata:
      annotations:
        'consul.hashicorp.com/connect-inject': 'true'
      labels:
        app: counting
    spec:
      containers:
      - name: counting
        image: hashicorp/counting-service:0.0.2
        ports:
        - containerPort: 9001
EOF

$ cat > counting.yaml <<EOFapiVersion: v1kind: ServiceAccountmetadata:  name: counting---apiVersion: v1kind: Servicemetadata:  name: countingspec:  selector:    app: counting  ports:  - port: 9001    targetPort: 9001---apiVersion: apps/v1kind: Deploymentmetadata:  labels:    app: counting  name: countingspec:  replicas: 1  selector:    matchLabels:      app: counting  template:    metadata:      annotations:        'consul.hashicorp.com/connect-inject': 'true'      labels:        app: counting    spec:      containers:      - name: counting        image: hashicorp/counting-service:0.0.2        ports:        - containerPort: 9001EOF

Create a deployment definition, service, and service account for the dashboard service named dashboard.yaml.

$ cat > dashboard.yaml <<EOF
apiVersion: v1
kind: ServiceAccount
metadata:
  name: dashboard
---
apiVersion: v1
kind: Service
metadata:
  name: dashboard
spec:
  selector:
    app: dashboard
  ports:
  - port: 9002
    targetPort: 9002
---
apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: dashboard
  name: dashboard
spec:
  replicas: 1
  selector:
    matchLabels:
      app: dashboard
  template:
    metadata:
      annotations:
        'consul.hashicorp.com/connect-inject': 'true'
        'consul.hashicorp.com/connect-service-upstreams': 'counting:9001'
      labels:
        app: dashboard
    spec:
      containers:
      - name: dashboard
        image: hashicorp/dashboard-service:0.0.4
        ports:
        - containerPort: 9002
        env:
        - name: COUNTING_SERVICE_URL
          value: 'http://localhost:9001'
EOF

$ cat > dashboard.yaml <<EOFapiVersion: v1kind: ServiceAccountmetadata:  name: dashboard---apiVersion: v1kind: Servicemetadata:  name: dashboardspec:  selector:    app: dashboard  ports:  - port: 9002    targetPort: 9002---apiVersion: apps/v1kind: Deploymentmetadata:  labels:    app: dashboard  name: dashboardspec:  replicas: 1  selector:    matchLabels:      app: dashboard  template:    metadata:      annotations:        'consul.hashicorp.com/connect-inject': 'true'        'consul.hashicorp.com/connect-service-upstreams': 'counting:9001'      labels:        app: dashboard    spec:      containers:      - name: dashboard        image: hashicorp/dashboard-service:0.0.4        ports:        - containerPort: 9002        env:        - name: COUNTING_SERVICE_URL          value: 'http://localhost:9001'EOF

Use kubectl to deploy the counting service.

$ kubectl apply -f counting.yaml
serviceaccount/counting created
service/counting created
deployment.apps/counting created

$ kubectl apply -f counting.yamlserviceaccount/counting createdservice/counting createddeployment.apps/counting created

Use kubectl to deploy the dashboard service.

$ kubectl apply -f dashboard.yaml
serviceaccount/dashboard created
service/dashboard created
deployment.apps/dashboard created

$ kubectl apply -f dashboard.yamlserviceaccount/dashboard createdservice/dashboard createddeployment.apps/dashboard created

To verify the services were deployed, refresh the Consul UI until you observe that the counting and dashboard services are running.

»Visit the dashboard

To visit the dashboard, forward the pod's port where the dashboard service is running to your local machine on the same port by providing the pod name (dashboard), which you specified in the service definition YAML file.

$ kubectl port-forward deploy/dashboard 9002:9002
Forwarding from 127.0.0.1:9002 -> 9002
Forwarding from [::1]:9002 -> 9002

$ kubectl port-forward deploy/dashboard 9002:9002Forwarding from 127.0.0.1:9002 -> 9002Forwarding from [::1]:9002 -> 9002

Visit localhost:9002 in your web browser. It will display the dashboard UI with a number retrieved from the counting service using Consul service discovery.

»Secure service communication with intentions

Consul intentions provide you the ability to control which services are allowed to communicate. Next, you will use intentions to test the communication between the dashboard and counting services.

»Create an intention that denies communication

You can use a Consul ServiceIntention CRD to create an intention that prevents the dashboard service from reaching its upstream counting service.

Create a file named deny.yaml that denies communication between the two services.

$ cat > deny.yaml <<EOF
apiVersion: consul.hashicorp.com/v1alpha1
kind: ServiceIntentions
metadata:
  name: dashboard-to-counting
spec:
  destination:
    name: counting
  sources:
    - name: dashboard
      action: deny
EOF

$ cat > deny.yaml <<EOFapiVersion: consul.hashicorp.com/v1alpha1kind: ServiceIntentionsmetadata:  name: dashboard-to-countingspec:  destination:    name: counting  sources:    - name: dashboard      action: denyEOF

Use kubectl to apply the intention.

$ kubectl apply -f deny.yaml
serviceintentions.consul.hashicorp.com/dashboard-to-counting created

$ kubectl apply -f deny.yamlserviceintentions.consul.hashicorp.com/dashboard-to-counting created

Verify the services are no longer allowed to communicate by returning to the dashboard UI. The service will display a message that the "Counting Service is Unreachable", and the count will display as "-1".

»Allow the application dashboard to communicate with the Counting service

Finally, remove the intention so that the services can communicate again.

$ kubectl delete -f deny.yaml
serviceintentions.consul.hashicorp.com "dashboard-to-counting" deleted

$ kubectl delete -f deny.yamlserviceintentions.consul.hashicorp.com "dashboard-to-counting" deleted

Intentions take effect rather quickly. The next time you visit the dashboard you'll notice that it's successfully communicating with the backend counting service again.

»Extend your knowledge

»Rolling updates to Consul

While running Consul you may want to make configuration and deployment updates. You can use helm upgrade to increase the number of agents, enable additional features, upgrade the Consul version, or change your configuration. For example, you could add a connectInject.default setting and set to true. When this setting is present and set to true, the injector will inject the Consul service mesh sidecars into all pods by default. Otherwise, pods must specify the injection annotation to opt-in to service mesh sidecar injection. If the default is set to true, pods can still use the same annotation to explicitly opt-out of injection.

You can practice using helm upgrade by updating your custom values file to enable sidecar injection by default.

In the text editor of your choice, modify your config.yaml file so that the connectInject stanza now includes a default: true setting.

connectInject:
  enabled: true
  default: true

connectInject:  enabled: true  default: true

Initiate the upgrade with the -f flag that passes in your new values file.

$ helm upgrade consul -f config.yaml hashicorp/consul
Release "consul" has been upgraded. Happy Helming!
NAME: consul
...TRUNCATED...
To learn more about the release, run:

  $ helm status consul
  $ helm get all consul

$ helm upgrade consul -f config.yaml hashicorp/consulRelease "consul" has been upgraded. Happy Helming!NAME: consul...TRUNCATED...To learn more about the release, run:
  $ helm status consul  $ helm get all consul

»Next steps

To learn more about Consul service mesh on Kubernetes, review the service mesh tutorials. To learn how to deploy Consul on a Kubernetes cluster, review the production deployment tutorial. To learn how to secure Consul and services for production, read the Secure Consul and Registered Services on Kubernetes tutorial.