Help & DocumentationSystem AdministrationDeployment & ConfigurationDeploying Metric InsightsMI Application Deployment on Google Cloud Kubernetes Engine (GKE)

MI Application Deployment on Google Cloud Kubernetes Engine (GKE)

This article describes how to deploy MI application on Google Cloud Kubernetes Engine (GKE).

Table of contents:

GKE Architecture Diagram for High Availability Deployment

  1. Create Infrastructure
    1. Create or Choose a Project
    2. Enable GKE API
    3. Create Kubernetes Cluster
    4. Create Kubernetes Cluster
    5. Configure gcloud and kubectl on your Local Machine
    6. Create Google Filestore Instance
    7. Create MySQL Database Instance
  2. Install MI Application

GKE Architecture Diagram for High Availability Deployment

MI Deployment on GKE.pdf

1. Create Infrastructure

1.1. Create or Choose a Project

In Google Cloud, related resources are organized in projects. A project is required to create a GKE cluster.

1.2. Enable GKE API

Access Google Cloud Platform > APIs & Services > Library

Search for 'Kubernetes Engine API', [Enable] if it has not been already enabled.

1.3. Create Kubernetes Cluster

Access Kubernetes Engine > Clusters

  1. [Create]
  2. Fill out cluster parameters
  3. Once you have configured your cluster settings, click [Create]

1.4. Configure gcloud and kubectl on your Local Machine

For detailed instructions based on your operating system, see the following documents:

 

  1. Add your cluster to kubectl:
gcloud container clusters get-credentials <Your Cluster> --zone <Cluster Zone>
  1. Verify the cluster is available:
kubectl get nodes

1.5. Create Google Filestore Instance

Access Storage menu > Filestore > Instances

  1. [Create Instance]
  2. Fill out the instance details, then [Create]
  3. Wait until the instance is created and note the following:
    • Assigned IP
    • Fileshare name that you have entered

1.6. Create MySQL Database Instance

Access Main menu > SQL

  1. [Create Instance]
  2. [Choose MySQL]
  3. Note down your root password
  4. Choose MySQL 8.0 as Database version
  5. Enable Private IP under connections
  6. Set the following Flag values:
    • log_bin_trust_function_creators: on
    • sql_mode: NO_ENGINE_SUBSTITUTION
    • require_secure_transport: OFF

For the recommended settings for other flags, refer to the following article:

NOTE: not all of these are available on Google SQL Instances. As of May 2022, the following flags can not be set:

  • skip-log-bin,
  • mysqlx,
  • key_buffer_size,
  • innodb_adaptive_hash_index,
  • connect_timeout,
  • innodb_open_files,
  • innodb_flush_method,
  • gtid_mode.
  1. [Create Instance], wait until the instance is created and note down the Private IP

2. Install MI Application

  1. Go to your MI installer folder, and generate Kubernetes manifest and related commands by running the following command:
./installer.py kubernetes --db-hostname <YOUR_MYSQL_PRIVATE_IP> --db-user root --db-password <YOUR_MYSQL_ROOT_PASSWORD> --load-balancer-type external --service-type LoadBalancer --storage-class nfs --nfs-server-address <YOUR FILESERVER_IP> --images-pull-secret-name metricinsights-docker-registry --dp-hostname metricinsights-dataprocessor --nfs-shared-folder <YOUR_FILESERVER_SHARE_NAME>  --namespace mi -o deployment.yml

This command will generate several Kubernetes commands, execute each one in sequence.

  1. Run the following command to get deployments.
kubectl get deployments -n mi

If everything is set up correctly, you should see output similar to the following:

  1. Find out the MI public IP:
kubectl get services -n mi

NOTE: See Set up Web SSL Certificate for Metric Insights.

  1. List pods available. Make sure all nodes are in a Ready state, and Status is Running. Otherwise, try again later:
kubectl get pods -n mi

Note the metricinsights-web-master pod name:

  1. Copy key and certificate to the pod
kubectl cp /path/to/server.crt <metricinsights-web-master>:/opt/mi/ssl/server.crt -n mi
kubectl cp /path/to/server.key <metricinsights-web-master>:/opt/mi/ssl/server.key -n mi
  1. Restart apache service
kubectl exec -ti -n mi <metricinsights-web-master> -- service apache2 restart
  1. Go to your site’s URL in a web browser to ensure that the SSL/TLS certificate is properly installed and working