Creating a GigaSpaces Cluster On Premise

 

gsctl is a simple CLI tool for creating GigaSpaces clusters. You can create a cluster in minutes with a single command – run java -jar gsctl.jar create.

In this topic, you will learn how to create, provision and install a GigaSpaces cluster on an on-premise machine, and then remove it by doing the following:

  • Create a GigaSpaces cluster.
  • Deploy GigaSpaces services in the cluster.
  • Undeploy the GigaSpaces services and tear down the cluster.

Prerequisites

Before beginning to work with the gsctl tool, you must do the following:

  • Verify that Java 8 or higher is installed on your local workstation.

  • Disable the password prompt when using sudo
  • Install Unzip on all the cluster machines
  • Ensure that you have a single pem file for all your on-premise machines
  • Verify connectivity to all the machines from your workstation
  • Centos machines only - disable the firewall (firewalld service) and make sure it is set to stay disabled on start (for example, when the machine reboots)

Downloading and Running the Utility

The gsctl tool can be downloaded from a dedicated repository. You can start an interactive shell for gsctl and run commands using short notation, or you can run commands from outside the interactive shell (from the OS shell) by adding java -jar gsctl.jar before each command.

You must run the utility from the same directory where you placed the gsctl.jar file.

To download the gsctl utility and launch the interactive shell:

  1. Create an empty directory on your machine. In this example, the drive is C: and the directory name is gsDemo.

  2. Download the most current version of the utility (gsctl.jar file) from the dedicated repository to the directory you created (gsDemo).

  3. To start the interactive shell, open a command window and type the following:

    java -jar gsctl.jar
    java -jar C:\gsDemo\gsctl.jar

Creating a GigaSpaces Cluster

Follow the steps in the procedure to create a GigaSpaces cluster and deploy the GigaSpaces services. All of the procedures on this page describe the short notation as used from inside the interactive shell.

To create a cluster:

  1. Create an empty directory on your machine. In this example, the directory name is gsDemo.

  2. Download the most current version of the gsctl.jar file from the dedicated repository to the directory you created (gsDemo) and extract the contents.
  3. Add the pem file to this directory.
  4. Open a command window and type the following to define a cluster called gs_demo_cluster:

    init  --mode=on-prem --cluster-name=gs_demo_cluster

    This command creates also a cluster.yaml file. Modify the yaml file as follows:

    • keyName - name of your pem file (without the .pem extension)

    • userName - name of the on-premise machine being used to create the cluster

    • profiles - the on-premise worker (client) name, along with the IP addresses of the host machines

      You must have 3 masters (gsManagers) and at least 1 worker.

    gsManagers: 3
    aws: null
    onPremise:
      keyName: "pem file"
      userName: "user name"
      masters:
        label: "GS Cluster [gs_demo_cluster] Master Group"
        profiles:
        - name: "default on premise master name"
          hosts:
          - "127.0.1.93"
          - "127.0.1.95"
          - "127.0.1.96"
      workers:
        label: "GS Cluster [gs_demo_cluster] Worker Group"
        profiles:
        - name: "default on premise worker name"
          hosts:
          - "127.0.1.97"
          - "127.0.1.98"
          - "127.0.1.99"
    name: "gs_demo_cluster"
  5. To create the cluster, run the following command:

    create

Deploying the GigaSpaces Services

After your cluster is up and running, you can deploy GigaSpaces services and microservices. The gsctl utility supports deploying stateful, stateless, and web services. The default service type is stateless.

The gsctl tool comes with sample feeder (stateful) and processor (stateless) services in the Services folder. You must use the default artifact repository when deploying the sample services.

For more information about the artifact repository, see the Managing the GigaSpaces Product Version topic.

To deploy the sample GigaSpaces services:

  1. When deploying the sample services, you should first deploy the processor service . As this is a stateful service, you need to specify both the service type and service name. Type the following command to deploy a processor service called mySpace:

    deploy --type=stateful mySpace data-processor.jar
  2. Next, deploy the feeder service. As this is a stateless service, you only have to specify the service name. Type the following command to deploy a feeder service called myFeeder:

    deploy --type=stateless myFeeder data-feeder.jar

Monitoring the GigaSpaces Services

After you deploy your GigaSpaces services and microservices, you can monitor them using the following built-in web interfaces:

  • Ops Manager
  • Grafana
  • Zipkin

To access the administration and monitoring tools:

  1. Run the following command:

    list-services

    This returns all the services with their URLs.

  2. Copy any of the GigaSpacesManager URLs into your browser to open Ops Manager and view the deployed GigaSpaces services.

  3. Copy the grafana URL to your browser to open Grafana and navigate to the pre-defined dashboards:

  4. Select the Telegraf system metrics dashboard in Grafana to view the cluster metrics:

Removing a GigaSpaces Cluster

You can delete your cluster when you no longer need it, in order to release the cloud resources. First, undeploy the GigaSpaces stateful service, and then tear down the cluster.

To remove the GigaSpaces cluster:

  1. In the directory where you deployed the GigaSpaces services, type the following command:

    undeploy mySpace

    The Space is undeployed from the cluster.

  2. In the directory where you created the cluster, run the following command:

    gsctl.jar destroy

    This tears down the cluster and deletes any dynamically created resources.