Scale out a CDP Basic Edition cluster
This topic describes the procedure for scaling out a Cloudera Data Platform (CDP) cluster.
Introduction to CDP scale-out
If your CDP cluster has insufficient compute or storage resources, you can scale it out horizontally. You can scale out nodes of the Core group type. By default, the new nodes have the same configuration as the existing nodes in the group.
Scaling out a CDP cluster incurs fees for ECS hardware and Alibaba Cloud Marketplace software orders.
After you pay for the scale-out order, you must add and deploy the nodes in the Cloudera Manager console.
Procedure
Log on to the Alibaba Cloud CDP console.
In the navigation pane, click the Cluster Management tab.
On the Cluster Management page, click Scale Out, or click the name of the target cluster.
On the cluster Details page, click Scale Out.
In the Scale Out dialog box, set Number of Instances to Add and Subscription Duration for the Core instance group.
Select the CDP terms of service, and then click Confirm.
After the payment is complete, the new scale-out nodes appear on the Host List page in about 5 minutes. Then, follow the instructions in the next section of this document to complete the scale-out operation in Cloudera Manager.
To generate a new scale-out order, you must first cancel the previous software and hardware orders for the instance in the Alibaba Cloud User Center console.
Add hosts in the Cloudera Manager console
Hosts managed by Cloudera Manager
After you complete the cluster scale-out operation in the Alibaba Cloud console, find the new instance group on the Host List page and note the node names for later use.
Log on to the Cloudera Manager console.
In the navigation pane, click Hosts > Add Hosts to open the Add Hosts wizard.
On the Add Hosts page, leave the default Add hosts to Cloudera Manager option selected and click Continue.
On the Specify Hosts page, enter the names of the new scale-out nodes that you noted in Step 1 into the Hostname field. Click Search to find the host information, and then click Continue.
On the Select Repository page, use the default custom repository for the cluster and click Continue.
On the Select JDK page, use the default options and click Continue.
On the Enter Login Credentials page, enter the root password for the cluster's ECS instances and click Continue.
On the Install Agents page, wait for the agents to be installed. This process takes about 3 minutes for a single-node scale-out.
Follow the remaining steps to finish adding the hosts to Cloudera Manager.
Add hosts to the CDP cluster and deploy services
After you add the hosts to Cloudera Manager, you must also add them to the CDP cluster by following these steps:
Log on to the Cloudera Manager console.
In the navigation pane, click Hosts > Add Hosts to open the Add Hosts wizard.
On the Add Hosts page, select Add hosts to Cluster and click Continue.
On the Specify Hosts page, select the hosts that you want to add to the cluster and click Continue.
On the Install Parcels page, wait for the parcels to be installed. This process takes about 4 to 5 minutes for a single node. You are then automatically taken to the Inspect Hosts page. Click Continue.
On the Select Host Template page, click Create to create a host template for the new nodes. For example, you can select HDFS DataNode, YARN NodeManager, and the gateways for each component. Enter a name for the template and click Create. You are returned to the Select Host Template page. Select the new template and click Continue.
On the Apply Host Template page, wait for the host template to be deployed and for the related roles to start. This process takes about 2 to 3 minutes. Click Continue.
On the Deploy Client Config page, wait for the client configurations to be deployed, and then click Finish.
After the scale-out is complete, some service components, such as Spark, may need to be restarted. A restart prompt appears on the cluster page in Cloudera Manager.