Terraform module to provision single or multiple member Hazelcast clusters on Kubernetes

This module creates the following resources required to provision Hazelcast

1. Deployment resource
2. Service resource

Architecture:

Prerequisites

1. Install Terraform from here
2. Allow terraform to access the kubernetes master using by following https://www.terraform.io/docs/providers/kubernetes/index.html

Usage

Note: Change the inputs to match your requirement.

module "hazelcast" {
  source            = "<module_path>"
  name              = "customer-orders"
  hazelcast_version = "3.12.7"
  cluster_members   = 3
  cpu               = "500m"
  memory            = "200Mi"
  labels = {
    created-by = "terraform"
    purpose    = "example"
  }
}

Examples

Try out the module functionality with the example defined https://github.com/everest-engineering/terraform-kubernetes-hazelcast/blob/master/examples/main.tf

1. Switch to examples directory cd examples
2. Initialize Terraform to download required plugins terraform init
3. Run plan to find out all resources that are going to be created terraform plan
4. Run apply to create those resources terraform apply
5. Deploy a hazelcast client pod and verify the connection. More details about this are here
6. Make sure to destroy them once you are done exploring using terraform destroy

Connecting to the Hazelcast cluster running inside Kubernetes

Hazelcast supports two strategies to simplify the deployment in a Kubernetes cluster. Official documentation is at https://github.com/hazelcast/hazelcast-kubernetes#understanding-discovery-modes

This module uses the DNS Lookup strategy to form a cluster. So, the client should also use this strategy to connect to the cluster.

Example Hazelcast java client project

Here is a sample test client written in java https://github.com/everest-engineering/terraform-kubernetes-hazelcast/blob/master/tests/hazelcast-java-client

As of now (on 23rd June 2020) only the java Hazelcast client implementation supports kubernetes discovery using the plugin https://github.com/hazelcast/hazelcast-kubernetes

You can verify the feature completeness for your favourite langauge at https://hazelcast.org/imdg/clients-languages/

Inputs

Name	Description	Type	Default	Required
cluster_members	number of members in the cluster	number	1	no
cpu	cpu units to request	string	n/a	yes
hazelcast_version	hazelcast version to deploy	string	n/a	yes
labels	labels to apply to all resources	map(string)	n/a	yes
memory	amount of memory to request	string	n/a	yes
name	name of the cluster	string	"hazelcast"	no
namespace	kubernetes namespace to deploy to	string	"default"	no

Outputs

Name	Description
service_dns	fully qualified dns name of the hazelcast service

Testing the module

Install minikube from here and start a kubernetes cluster in local.

Setup

1. Test setup is automated in setup.sh. The shell script does the following tasks.
   • deploys the example hazelcast cluster
   • verifies that the cluster has formed successfully
   • builds the client project jar
   • builds a docker image using the above jar in Minikube
   • deploys a kubernetes job to run the client in the cluster
   • verifies that client has connected to the cluster successfully

Tear down

1. Use the script teardown.sh to remove all test resources

Contributing

We appreciate your help!

Open an issue or submit a pull request for an enhancement. Browse through the current open issues.

License

Authors