Run SQLFlow Playground on Google Cloud Platform

Step 0: Get a k8s cluster.

Step 1: Grant pod-related permissions for the default user.

$ kubectl apply -f manifests/rbac.yaml

Step 2: Deploy the JupyterHub pod on Kubernetes.

You need a GitHub OAuth application and fill in the GITHUB_CLIENT_ID, GITHUB_CLIENT_SECRET and OAUTH_CALLBACK_URL of the sqlflow-jhub.yaml, or you can contact the administrator to get these credential.

$ kubectl create -f manifests/sqlflow-jhub.yaml
deployment.apps/sqlflow-jhub created
service/sqlflow-jhub created

Step 3: Find the <EXTERNAL-IP> for service/sqlflow-jhub. In this case, it is 34.67.182.237.

$ kubectl get all
NAME                                READY   STATUS    RESTARTS   AGE
pod/sqlflow-jhub-77858f4655-7wpt8   1/1     Running   0          2m39s

NAME                   TYPE           CLUSTER-IP      EXTERNAL-IP     PORT(S)          AGE
service/kubernetes     ClusterIP      10.47.240.1     <none>          443/TCP          80m
service/sqlflow-jhub   LoadBalancer   10.47.249.179   34.67.182.237   8000:31721/TCP   2m39s

NAME                           DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
deployment.apps/sqlflow-jhub   1         1         1            1           2m39s

NAME                                      DESIRED   CURRENT   READY   AGE
replicaset.apps/sqlflow-jhub-77858f4655   1         1         1       2m40s

Step 4: Append a host record in your /etc/hosts file:

<EXTERNAL-IP> play.sqlflow.org

Step 5: Open a browser, go to play.sqlflow.org:8000 and login using your Github account.