Alfresco Content Services Helm Deployment on KinD cluster
This page describe how to get ACS up and running on a Kubernetes In Docker (KinD) cluster.
KinD is better suited to run helm workloads inside CI environments but works well also for local development because it just requires a working Docker server.
Prerequisites
- Docker installed on your machine (Docker Desktop, Rancher Desktop, Podman and similar)
- Kubernetes CLI (kubectl) installed on your machine
- Helm CLI installed on your machine
Step 1: Install Kind
Take a look to the KinD quickstart to learn how to install the binary cli on your machine and to learn briefly the main commands that you can run.
Step 2: Create a Kind Cluster
Run the following command to create the default Kind cluster:
kind create cluster
You can also create a cluster targeting a specific Kubernetes version, for example:
kind create cluster --image kindest/node:v1.34.3@sha256:08497ee19eace7b4b5348db5c6a1591d7752b164530a36f855cb0f2bdcbadd48
The node image ref needs to be retrieved from the KinD releases page, by looking at the release notes of your current kind version (run kind version to check your version).
Wait for the Kind cluster to be created. This usually takes a few minutes.
Step 3: Install cloud-provider-kind
Install the cloud-provider-kind plugin to enable cloud provider features on your KinD cluster.
You can find the latest binaries available on the GitHub releases page of the project.
If you are using brew, you can install it with:
brew install cloud-provider-kind
Step 4: Run cloud-provider-kind
cloud-provider-kind is a separate binary that you need to keep running in the background while your cluster is up and running.
It will automatically attach to your KinD cluster and monitor for any Service of type LoadBalancer that you create, by assigning them an IP address and making them reachable from your local machine.
Open a new terminal and run the following command:
sudo cloud-provider-kind
If you are running a Docker server that runs within a dedicated VM (like Podman Desktop on Mac and Windows), it will automatically detect it and enable the --enable-lb-port-mapping option to make the LoadBalancer endpoint easily reachable from your local machine, exposed on localhost.
This is not strictly required when running on native Linux, but accessing Alfresco via localhost is usually more compatible when using plain http access.
Step 5: Install an Ingress Controller
See Traefik section.
The ingress-nginx section is kept for reference only, as ingress-nginx is deprecated and not recommended for new deployments.
Once traefik is installed, verify that the LoadBalancer external IP has been assigned by running:
kubectl get svc -n traefik
You should see something like:
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
traefik LoadBalancer 10.96.209.229 10.89.0.3 80:30539/TCP,443:31996/TCP 4m
The external-ip is the IP address of the container that cloud-provider-kind plugin has created to route the traffic to the Traefik controller, and that you can see when looking at the running containers in your Docker server:
docker ps
You will see a container named kindccm-... with the ports 80 and 443 exposed, which is the one routing the traffic to Traefik, making it reachable from your local machine.
xxxxxxxxxxxx docker.io/envoyproxy/envoy:v1.33.2 ... 0.0.0.0:36059->80/tcp, 0.0.0.0:36315->443/tcp, 0.0.0.0:39449->10000/tcp kindccm-xxxxxxxxx
xxxxxxxxxxxx docker.io/envoyproxy/envoy:v1.33.2 ... 0.0.0.0:43741->80/tcp, 0.0.0.0:36821->10000/tcp kindccm-gw-xxxxxxxxx
Try accessing the port that exposes port 80, e.g. localhost:36059, from your browser and you should see the default 404 page not found of Traefik, which means that the traffic is correctly routed to the Traefik controller.
Also verify that the traefik pod logs show the incoming request:
kubectl logs -n traefik -l app.kubernetes.io/name=traefik
Step 6: Install metrics server
Optionally, you can install metrics server to gather metrics from node and pods.
Make sure to enable insecure tls option otherwise metrics collection would not work under KinD:
kubectl patch -n kube-system deployment metrics-server --type=json \
-p '[{"op":"add","path":"/spec/template/spec/containers/0/args/-","value":"--kubelet-insecure-tls"}]'
Wait for a few minutes and then try to run:
kubectl top node
Upon installing ACS, you can check current cpu and memory usage via:
kubectl top pods
Conclusion
Now that you have successfully set up a Kind cluster with an Ingress controller and metrics-server, you can now proceed with installing ACS via helm charts as per Desktop deployment.