Google Kubernetes Engine (GKE)

• Helm chart
• YAML (remote disk)
• YAML (local disk)

Prerequisites

You must have a GKE cluster that has Helm configured. If you have not installed the Helm client (helm), see Installing Helm.

The YugabyteDB Helm Chart has been tested with the following software versions:

• GKE running Kubernetes 1.18 (or later) with nodes such that a total of 12 CPU cores and 45 GB RAM can be allocated to YugabyteDB. This can be three nodes with 4 CPU core and 15 GB RAM allocated to YugabyteDB. n1-standard-8 is the minimum instance type that meets these criteria.
• Helm 3.4 or later
• Docker image for YugabyteDB (yugabytedb/yugabyte) 2.1.0 or later
• For optimal performance, ensure you set the appropriate system limits using ulimit on each node in your Kubernetes cluster.

The following steps show how to meet these prerequisites.

• Download and install the Google Cloud SDK.

• Configure defaults for gcloud

Set the project ID as yugabyte. You can change this as needed.

Set the default compute zone as us-west1-b. You can change this as needed.

• Install kubectl

After installing Cloud SDK, install the kubectl command line tool by running the following command.

Note that GKE is usually 2 or 3 major releases behind the upstream/OSS Kubernetes release. This means you have to make sure that you have the latest kubectl version that is compatible across different Kubernetes distributions if that's what you intend to.

• Ensure helm is installed

First, check to see if Helm is installed by using the Helm version command.

You should see something similar to the following output. Note that the tiller server side component has been removed in Helm 3.

version.BuildInfo{Version:"v3.0.3", GitCommit:"ac925eb7279f4a6955df663a0128044a8a6b7593", GitTreeState:"clean", GoVersion:"go1.13.6"}

1. Create a GKE cluster

Create a Kubernetes cluster, if you have not already done so, by running the following command.

As stated in Prerequisites above, the default configuration in the YugabyteDB Helm Chart requires Kubernetes nodes to have a total of 12 CPU cores and 45 GB RAM allocated to YugabyteDB. This can be three nodes with 4 CPU cores and 15 GB RAM allocated to YugabyteDB. The smallest Google Cloud machine type that meets this requirement is n1-standard-8 which has 8 CPU cores and 30GB RAM.

2. Create a YugabyteDB cluster

Add charts repository

To add the YugabyteDB charts repository, run the following command.

Fetch updates from the repository

Make sure that you have the latest updates to the repository by running the following command.

Validate the Chart version

Output:

NAME                    CHART VERSION   APP VERSION     DESCRIPTION
yugabytedb/yugabyte     2.13.0           2.13.0.0-b42    YugabyteDB is the high-performance distributed ...

Install YugabyteDB

Run the following commands to create a namespace and then install Yugabyte.

Check the cluster status

You can check the status of the cluster using various commands noted below.

Output:

NAME: yb-demo
LAST DEPLOYED: Thu Feb 13 13:29:13 2020
NAMESPACE: yb-demo
STATUS: deployed
REVISION: 1
TEST SUITE: None
NOTES:
1. Get YugabyteDB Pods by running this command:
  kubectl --namespace yb-demo get pods

2. Get list of YugabyteDB services that are running:
  kubectl --namespace yb-demo get services

3. Get information about the load balancer services:
  kubectl get svc --namespace yb-demo

4. Connect to one of the tablet server:
  kubectl exec --namespace yb-demo -it yb-tserver-0 -- bash

5. Run YSQL shell from inside of a tablet server:
  kubectl exec --namespace yb-demo -it yb-tserver-0 -- ysqlsh -h yb-tserver-0.yb-tservers.yb-demo

6. Cleanup YugabyteDB Pods
  helm delete yb-demo --purge
  NOTE: You need to manually delete the persistent volume
  kubectl delete pvc --namespace yb-demo -l app=yb-master
  kubectl delete pvc --namespace yb-demo -l app=yb-tserver

Check the pods.

NAME           READY     STATUS    RESTARTS   AGE
yb-master-0    1/1       Running   0          4m
yb-master-1    1/1       Running   0          4m
yb-master-2    1/1       Running   0          4m
yb-tserver-0   1/1       Running   0          4m
yb-tserver-1   1/1       Running   0          4m
yb-tserver-2   1/1       Running   0          4m

Check the services.

NAME                 TYPE           CLUSTER-IP      EXTERNAL-IP    PORT(S)                                        AGE
yb-master-ui         LoadBalancer   10.109.39.242   35.225.153.213 7000:31920/TCP                                 10s
yb-masters           ClusterIP      None            <none>         7100/TCP,7000/TCP                              10s
yb-tserver-service   LoadBalancer   10.98.36.163    35.225.153.214 6379:30929/TCP,9042:30975/TCP,5433:30048/TCP   10s
yb-tservers          ClusterIP      None            <none>         7100/TCP,9000/TCP,6379/TCP,9042/TCP,5433/TCP   10s

You can even check the history of the yb-demo deployment.

Output:

REVISION  UPDATED                   STATUS    CHART           APP VERSION   DESCRIPTION
1         Tue Apr 21 17:29:01 2020  deployed  yugabyte-2.13.0  2.13.0.0-b42  Install complete

Connect using YugabyteDB shells

To connect and use the YSQL Shell ysqlsh, run the following command.

To connect and use the YCQL Shell ycqlsh, run the following command.

Connect using external clients

To connect an external program, get the load balancer EXTERNAL-IP IP address of the yb-tserver-service service and connect to the 5433 / 9042 ports for YSQL / YCQL services respectively.

NAME                 TYPE           CLUSTER-IP      EXTERNAL-IP   PORT(S)                                        AGE
...
yb-tserver-service   LoadBalancer   10.98.36.163    35.225.153.214     6379:30929/TCP,9042:30975/TCP,5433:30048/TCP   10s
...

Configure cluster

You can configure the cluster using the same commands and options as Open Source Kubernetes.

Independent LoadBalancers

By default, the YugabyteDB Helm Chart will expose the client API endpoints as well as master UI endpoint using two LoadBalancers. If you want to expose the client APIs using independent LoadBalancers, you can do the following.

You can also bring up an internal LoadBalancer (for either YB-Master or YB-TServer services), if required. Just specify the annotation required for your cloud provider. The following command brings up an internal LoadBalancer for the YB-TServer service in Google Cloud Platform.