Deploy Platform Operator via GKE and Expose LB for External Clients

Owned by Aaron Franz
Aug 30, 2024
3 min read

Deploying Platform Operator and Platform on GKE

  • Create cluster via GKE GUI

image-20240830-122439.png

  • These settings can be configured as necessary to fit your environment.

  • One deployed, select ‘Connect’ to retrieve the CLI string required for kubectl connectivity.

image-20240830-122626.png

  • Deploy operator and platform. Note that if using a local, custom helm chart for the HiveMQ platform, the path to the local chart can be used in place of hivemq/hivemq-platform, with the -f option. helm install -f local/file.

    helm install test-operator hivemq/hivemq-platform-operator --wait && helm install test-platform hivemq/hivemq-platform --wait

  • If you encounter an error citing that Break is not defined, ensure that your installed Helm version is at least version 1.12.

    Error: INSTALLATION FAILED: parse error at (hivemq-platform/templates/_helpers.tpl:93): function "break" not defined

  • Once deployed successfully, validation of success can be done with kubectl describe pods. This will provide the listed pods and validate their status.

 

GKE Ingresses, as created through the GUI, only support HTTP/HTTPS connections. In order to allow access to the cluster from external clients, a separate load balancer will need to be created.

  • Using the default deployment with the platform operator and platform pods, access your google cloud console, and perform the following :

touch lb-values.yaml

  • This will create a new, empty lb-values.yaml file in your current directory. Using your preferred editor, change the values of this lb-values.yaml file to the following :

  • Note that the selector field here contains a tag hivemq-platform: your-platform-deployment-name. The value section ‘your-platform-deployment-name’ here will need to be replaced by the name provided to the HiveMQ platform during the helm install command.

  • This can also be confirmed by running kubectl get pods to list pod names, then kubectl describe pod POD_NAME. This will provide a description of the provided pod name. Within the labels, and annotations section, you will find the following :

  • We are specifically looking for the hivemq-platform=test-platform line here. This will be the name provided during the helm install, and is used to point the new load balancer to these pods.

  • Once the lb-values.yaml file has been updated with the appropriate selector value, save the file and create this service with kubectl apply -f lb-values.yaml.

  • This will begin to automatically create the service from this file, and a Google Cloud Controller will be awoken to automatically provide an external address. This can be seen by entering kubectl get services.