GCP-WEKA deployment Terraform package description

WEKA provides a ready-to-deploy GCP-WEKA deployment Terraform package that you can customize to install the WEKA cluster on GCP.

The Terraform package contains the following modules:

• setup_network: includes vpcs, subnets, peering, firewall, and health check.

• service_account: includes the service account used for deployment with all necessary permissions.

• deploy_weka: includes the actual WEKA deployment, instance template, cloud functions, workflows, job schedulers, secret manager, buckets, and health check.

• shared_vpcs (optional): includes VPC sharing the WEKA deployment network with another hosting project. For example, when deploying a private network.

GCP-WEKA deployment Terraform package supported types

The Terraform package supports the following deployment types:

• Public cloud deployments: Require passing the get.weka.io token to Terraform for downloading the WEKA release from the public get.weka.io service. The following examples are provided:

  • Public VPC

  • Public VPC with creating a worker pool

  • Public VPC with an existing public network

  • Public VPC with multiple clusters

  • Public VPC with a shared VPC

  • Public VPC with an existing worker pool and VPC

• Private cloud deployments: Require uploading the WEKA release tar file into the yum repository (instances can download the WEKA release from this yum repository). The following examples are provided:

  • Private VPC with creating a worker pool

  • Private VPC with an existing network

  • Private VPC with an existing worker pool and VPC

  • Private VPC with multiple clusters

  • Private VPC with a shared VPC

Terraform example

The following is a basic example in which you provide the minimum detail of your cluster, and the Terraform module completes the remaining required resources, such as cluster size, machine type, and networking parameters.

You can use this example as a reference to create the main.tf file.

provider "google" {
  region  = "europe-west1"
  project = "PROJECT_ID"
}

module "weka_deployment" {
  source                         = "weka/weka/gcp"
  version                        = "4.0.0"
  cluster_name                   = "my_cluster_name"
  project_id                     = "PROJECT_ID"
  prefix                         = "my_prefix"
  region                         = "europe-west1"
  zone                           = "europe-west1-b"
  cluster_size                   = 7
  nvmes_number                   = 2
  get_weka_io_token              = "getwekatoken"
  machine_type                   = "c2-standard-8"
  subnets_range                  = ["10.222.0.0/24", "10.223.0.0/24", "10.224.0.0/24", "10.225.0.0/24"]
  allow_ssh_cidrs                = ["0.0.0.0/0"]
  allow_weka_api_cidrs           = ["0.0.0.0/0"]

}
output "weka_cluster" {
  value = module.weka_deployment
}

For the descriptions of the parameters, refer to the GCP-WEKA deployment Terraform package.

Private network considerations

To deploy a private network, the parameter private_network = true on the setup_network and deploy_weka modules level.

Depending on the required network topology, the following parameters are optional for private networking:

• To download the WEKA release from a local bucket, set the local bucket location in the install_url parameter on the deploy_weka module level.

• For Centos7 only, a distributive repository is required to download kernel headers and additional build software. To auto-configure yum to use a distributive repository, run yum_repo_server.

• If a custom image is required, use weka_image_id.

Object store integration

The Terraform package can automate the addition of a Google Cloud Storage bucket for use as object storage.

Procedure

1. In the main.tf file, add the following fields:

   • tiering_enable_obs_integration: Set the value to true.

   • tiering_obs_name: Match the value to an existing bucket in Google Cloud Storage.

   • tiering_ssd_percent: Set the percentage to your desired value.

Example:

tiering_enable_obs_integration=true 
tiering_obs_name="myBucketName"
tiering_ssd_percent=20

If you do not specify the name of an existing bucket using tiering_obs_name but specify tiering_enable_obs_integration=true then a new Cloud Storage bucket is created automatically. The name format of the new bucket is: <project_id>-<prefix>-<cluster_name>-obs