Start Kubernetes prepared for huge data handling

Building a Custom Kubernetes Cluster with Kind and HariKube

This tutorial walks you through setting up a Kubernetes cluster using Kind (Kubernetes in Docker), but with a twist: you’ll use custom Kubernetes images and an external HariKube middleware for storage. This approach is ideal for developers and operators who want to test new Kubernetes versions or integrate with a different storage backend.

*Large dataset support based on storage-level filtering is not available in vanilla Kubernetes. Follow this post for instructions to running a custom version of Kubernetes.

Running HariKube Middleware

First, let’s set up your HariKube data store. Normally, Kubernetes relies on ETCD as its primary database. Instead of letting Kind run its own embedded ETCD instance, you’ll launch a custom “middleware” container that provides an ETCD API endpoint—powered by HariKube.

HariKube images aren’t public yet. If you’d like to try them, request a trial version on the Get Started page.

Start by authenticating your local Docker client with the private registry at registry.harikube.info. This step is essential for pulling images from the registry.

1

docker login registry.harikube.info

Now, run the HariKube middleware container:

1
2
3
4
5
6
7

docker run -d --name harikube_middleware \
  -e TOPOLOGY_CONFIG=secret://default/topology-config \
  -e ENABLE_TELEMETRY_PUSH=true \
  -v harikube_db:/db \
  -p 2369:2379 \
  registry.harikube.info/harikube/middleware:beta-v1.0.0-13 \
  --endpoint='multi://sqlite:///db/main.db?_journal=WAL&cache=shared'

Configuring and Creating the Kind Cluster

Next, let’s configure Kind to use your custom HariKube backend. The kind-config.yml file tells Kind how to build the cluster and connect to your middleware.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17

kind: Cluster
apiVersion: kind.x-k8s.io/v1alpha4
featureGates:
  "CustomResourceFieldSelectors": true
  "WatchList": true
  "WatchListClient": true
nodes:
- role: control-plane
  image: kindest/node:v1.33.1
  kubeadmConfigPatches:
  - |
    kind: ClusterConfiguration
    apiServer:
        extraArgs:
          watch-cache: "false"
          storage-media-type: application/json
          etcd-servers: "http://172.17.0.1:2369"

To spin up your cluster, run:

1

kind create cluster --name harikube-cluster --config kind-config.yml

This command uses your configuration file to create a new Kubernetes cluster named harikube-cluster, with HariKube as the storage backend.

Accessing Custom Kubernetes Images (Optional, but Recommended)

If you want true large dataset support, you’ll need to use HariKube’s pre-built Kubernetes images from our private registry. This step is optional for basic setups, but highly recommended for production or heavy workloads.

First, pull the custom images:

1
2

docker pull registry.harikube.info/harikube/kube-apiserver-amd64:v1.33.1
docker pull registry.harikube.info/harikube/kube-controller-manager-amd64:v1.33.1

Then, load the images into your Kind node:

1
2

kind load docker-image -n harikube-cluster registry.harikube.info/harikube/kube-apiserver-amd64:v1.33.1
kind load docker-image -n harikube-cluster registry.harikube.info/harikube/kube-controller-manager-amd64:v1.33.1

Finally, update the component manifests to use your custom images:

1
2

docker exec harikube-cluster-control-plane sed -i 's|registry.k8s.io/kube-apiserver:v1.33.1|registry.harikube.info/harikube/kube-apiserver-amd64:v1.33.1|' /etc/kubernetes/manifests/kube-apiserver.yaml
docker exec harikube-cluster-control-plane sed -i 's|registry.k8s.io/kube-controller-manager:v1.33.1|registry.harikube.info/harikube/kube-controller-manager-amd64:v1.33.1|' /etc/kubernetes/manifests/kube-controller-manager.yaml

Wait for your Kubernetes cluster to restart and come back online with the custom images. You’re now ready to handle huge datasets with confidence!

Ready for the next step? Learn how to deploy the HariKube operator to manage dynamic database topology. →

That’s it! You’ve just built a Kubernetes cluster that’s ready for serious data handling. With HariKube, you can design your own data topology and take advantage of lower latency, higher throughput, true data isolation, virtually unlimited storage, and a much simpler development experience. HariKube supports both flat and hierarchical topologies, so you can organize your databases like leaves on a tree.

Thank you for following along! If you have questions or ideas, please share them—we’d love to hear from you.