The samples in this directory illustrate simple Helm charts that can be used to deploy ClickHouse and ClickHouse Keeper. THEY ARE SAMPLES ONLY AND NOT GUARANTEED TO WORK. USE OFFICIAL ALTINITY HELM CHARTS INSTEAD.

We do not make an effort to show everything users can do with Helm. The samples provide a starting point for further embroidery.

Important notes:

1. ClickHouse Keeper resources (CHK) are not stable yet. They should not be used for production.
2. Connections are unencrypted. Do not use them for sensitive data.
3. If you select the LoadBalancer service type in values.yaml it will open one or more ports to the Internet.

bfe1c81 (Teach helm not to create external load balancer on EKS)

You'll need to do the following to get started.

• Get administrative access to Kubernetes. For testing clickhouse-hello, Minikube will do.
• Install kubectl
• Install helm
• Use helm to install the clickhouse-operator using the commands shown below.

helm repo add clickhouse-operator https://docs.altinity.com/clickhouse-operator/
helm install clickhouse-operator clickhouse-operator/altinity-clickhouse-operator

• Use helm to install ClickHouse Keeper.

helm repo add kubernetes-blueprints https://docs.altinity.com/kubernetes-blueprints-for-clickhouse
helm install keeper kubernetes-blueprints/keeper-sts

Please note the Altinity Operator project instructions regarding operator upgrade with Helm. You need to run the custom resource definition files independently.

This shows how to get started with the 'hello' chart. Examples use defaults from the values.yaml file.

Clone this repo and install the Helm clickhouse-hello chart directly from the file system.

git clone https://github.com/Altinity/clickhouse-sql-examples.git
cd clickhouse-sql-examples/helm
helm install hello clickhouse-hello

If you would like to see what deviltry is afoot before installing, run the following to see the manifests.

helm install --debug --dry-run test clickhouse-hello

Here's how to connect and make sure ClickHouse can see the Keeper server.

kubectl exec -it chi-hello-ch-0-0-0 -- clickhouse-client --user=helm-user --password=topsecret
SELECT * FROM system.zookeeper WHERE path = '/'

If things are correctly wired, both commands will succeed and you'll see a list of path names in Keeper.

If you would like to connect to the server from outside, use kubectl port forward as shown below.

kubectl port-forward service clickhouse-hello 9000:9000 &
clickhouse-client --user=helm-user --password=topsecret

The login and password are set in values.yaml. Current values are stored in Secret clickhouse-hello. If you forget the login and password you can find them at any time using the following command:

kubectl get secret/clickhouse-hello -o json | jq '.data' | jq 'map_values(@base64d)'

You may need to install 'jq' for this to work.

You can change the settings in values.yaml and reapply the chart.

helm upgrade hello clickhouse-hello

You can also override using your own values.

helm upgrade hello clickhouse-hello -f newvalues.yaml

Remove the sample resources using helm uninstall. You'll also have to delete PVCs explicitly. This is a safety precaution to prevent accidental deletion of clusters and their storage.

helm uninstall hello
kubectl delete pvc -l application_group=hello

The clickhouse-aws chart shows how to run ClickHouse on AWS EKS across AZs using a nodeSelector to pin resources to run on specific VMs types. See the local README.md for more information.

Please log issues on this project and/or submit PRs if you have solutions to them.