Deploying to Kubernetes
In order to deploy this to a Kubernetes cluster, we will first need to create a container image for the application.
Later on, we will also show how you can build this in the cloud, and automatically re-deploy when you make changes.
Creating a container
You can build container images locally. First, ensure that you have an up-to-date Maven build:
mvn clean packageAnd then, create a container image from it:
docker build -t my-quarkus-mqtt-starter . -f src/main/docker/Dockerfile.jvm
Of course, you can simply replace the docker command with podman, it works the same way.
|
Pushing images
The previous step created a container image in your local machine. However, when using a Kubernetes cluster, that may not be sufficient. You might need to push the image to a container registry. Reasonable choices are quay.io or GitHub container registry.
Assuming you push to quay.io with your user being myaccount, you would need to run the following commands:
docker tag my-quarkus-starter quay.io/myaccount/my-quarkus-mqtt-starter:latest
docker push quay.io/myaccount/my-quarkus-mqtt-starter:latestDeploying the application
In the folder deploy/, you will find
some example deployment files.
First, if you want to deploy your changed container image, you will need to replace the image name in the file
deploy/020-deployment.yaml:
kind: Deployment
apiVersion: apps/v1
metadata:
name: quarkus-mqtt-integration-starter
# …
spec:
# …
template:
# …
spec:
containers:
- name: service
image: ghcr.io/drogue-iot/quarkus-mqtt-integration-starter-jvm:latest (1)
imagePullPolicy: Always
# …| 1 | The original name image, which you may replace if you want to deploy your own version |
| For all examples, it is a good idea to use a dedicated namespace, in order to isolate your deployment. |
OpenShift
If you are using OpenShift as a cluster, you can simply run the following command to deploy the application:
kubectl apply -f deploy/You can then get the HTTP endpoint using the following command:
kubectl get route quarkus-mqtt-integration-starterWhich should print out the information on the console like this:
NAME HOST/PORT PATH SERVICES PORT TERMINATION WILDCARD
quarkus-mqtt-integration-starter quarkus-mqtt-integration-starter-my-user-stage.apps.sandbox-m2.ll9k.p1.openshiftapps.com quarkus-mqtt-integration-starter http edge/Redirect NoneThe hostname is in the "HOST" column, and normally it is an HTTPS link. So the web-frontend URL (in this case) would be https://quarkus-mqtt-integration-starter-my-user-stage.apps.sandbox-m2.ll9k.p1.openshiftapps.com
Using OpenShift Developer Sandbox
You can also use the Developer Sandbox for Red Hat OpenShift to directly host this application.
Azure Kubernetes Service
Before you start, you will need an Azure account and have the az command installed and connected to your account.
|
This may cost money
Deploying the example on Azure may result on costs being billed to your account. You need to understand what the following commands actually do and ensure that you clean up resources after you are finished testing. If in doubt, consider a locally hosted Kubernetes solution like Minikube or Kind. Also see the documentation: https://docs.microsoft.com/en-us/azure/aks/ |
Creating a cluster
If you don’t have a cluster yet, you can create a simple test cluster by executing the following commands.
|
The following steps are an opinionated setup, focused on this specific use case. It follows the quick start instructions for the Azure Kubernetes Service. Also see the documentation: https://docs.microsoft.com/en-us/azure/aks/kubernetes-walkthrough |
az group create --name my-quarkus-mqtt-starter --location eastus
az aks create --resource-group my-quarkus-mqtt-starter --name my-quarkus-mqtt-cluster --node-count 1 --enable-addons monitoring,http_application_routing --generate-ssh-keys
az aks get-credentials --resource-group my-quarkus-mqtt-starter --name my-quarkus-mqtt-clusterFinally, verify that the cluster is running and that kubectl is properly configured to access it. Execute the following
command:
kubectl get nodesWhich should result in an output like this:
NAME STATUS ROLES AGE VERSION
aks-nodepool1-20363394-vmss000000 Ready agent 81s v1.19.11Deploying the application
Create a new namespace:
kubectl create namespace quarkus-mqtt-starterDeploy the application:
kubectl -n quarkus-mqtt-starter apply -f deploy/010-configuration.yaml
kubectl -n quarkus-mqtt-starter apply -f deploy/020-deployment.yaml
kubectl -n quarkus-mqtt-starter apply -f deploy/030-service.yamlRouting ingress traffic
az aks show --resource-group my-quarkus-mqtt-starter --name my-quarkus-mqtt-cluster --query addonProfiles.httpApplicationRouting.config.HTTPApplicationRoutingZoneName -o tableThis should print out the cluster’s domain name:
Result
-------------------------------------
d578cb0564094cf8979c.eastus.aksapp.io (1)| 1 | Cluster DNS name |
Create a new ingress using:
kubectl apply -f - <<EOF
kind: Ingress
apiVersion: networking.k8s.io/v1
metadata:
namespace: quarkus-mqtt-starter
name: web-frontend
annotations:
kubernetes.io/ingress.class: addon-http-application-routing
spec:
rules:
- host: web-frontend.d578cb0564094cf8979c.eastus.aksapp.io (1)
http:
paths:
- path: /
pathType: Prefix
backend:
service:
name: quarkus-mqtt-integration-starter
port:
name: http
EOF| 1 | A combination of web-frontend. plus the cluster domain name from before. |
The link to the web-frontend is (in this case): http://web-frontend.d578cb0564094cf8979c.eastus.aksapp.io
Clean up
If you created a cluster in an earlier step, you should clean it up by deleting the resource group.
| The following command will delete the whole resource group without asking for confirmation. Ensure that this is really what you want to do. |
az group delete --name my-quarkus-mqtt-starter --yes| To avoid unexpected charges, manually double check that all resources have indeed been cleaned up. |