The quickest way to launch a cluster with Cloudbreak is through the CLI. This allows you to specify all of the required settings and submit the request to Cloudbreak. The cluster will show up in Cloudbreak just like it does when launched via the wizard. The quickest way to generate the JSON template is to set up a cluster through the wizard.
Generating the JSON Template from Cloudbreak GUI
The first way you can create the template is from the Create Cluster wizard. In the Cloudbreak GUI, walk through all of the steps to configure a cluster. At the end, before you click "Create Cluster" you can click "Show CLI Command".
The second way is to go the a running cluster in the Cloudbreak GUI. Click the "Actions" button and you can click "Show CLI Command".
The CLI command dialog comes up with the JSON template in the top text box and the Cloudbreak CLI command in the lower text box. Click the "Copy the JSON" button and the paste it into a text file. Copy the CLI command to a note for later use.
Preparing the JSON template for creating a cluster
The JSON template does not have the sensitive values that were used in creating the cluster and they will have to be filled in. Most notably, the password in the ambari section. In the copied JSON, it will be blank, but you must set it before a cluster can be created with the JSON file.
"password": "SuperSecretL0ngPassword!!one",
You may also want to change input values for variables in a dynamic blueprint. The HA blueprint discussed in "Cloudbreak blueprint for high-availability Hadoop and Hive" is a dynamic blueprint, allowing you to change the value of dfs.namespace for each cluster you create. Set this in the inputs section. In this example, it is set to "hdp301".
"inputs": {
"dfs.nameservices": "hdp301"
},
Many of the configurations for HA in blueprints do not pass validation and therefore require a setting in the blueprint section. If so you will see an error when you try to create a cluster from the CLI. This could be for HIVE_SERVER, NAMENODE or other components.
ERROR: status code: 500, message: Incorrect number of 'HIVE_SERVER' components are in '[master_nn1, master_nn2]' hostgroups: count: 2, min: 1 max: 1
If you get an error, you will have to add "validateBlueprint": false, in the ambari section where the blueprint name is specified:
"ambari": {
"blueprintName": "hdp30-ha",
"validateBlueprint": false,
"userName": "admin",
Otherwise you may get a validation error when you
Creating the cluster with the CLI
Once you have the JSON template ready, you can create a cluster by just running the 'cb cluster create' command. This will return quickly with a successful return code and you will see the cluster in the Cloudbreak GUI. The cluster will be built just as it is when you click create cluster in the GUI wizard.
$ ../bin/cb cluster create --cli-input-json ../clusters/cb-hdp301-ha.json --name hdp301-ha $ echo $? 0
Changing settings in the JSON file
Now that you have a JSON template, you will be able to see where you can change configurations such as the instanceType or nodeCount for each host group, the HDP version and the repo paths to customize each new cluster you create.
