PAI Designer nodes - DataWorks - Alibaba Cloud Documentation Center

In DataWorks, you can use PAI Designer nodes to load PAI Designer pipelines and configure scheduling settings to run them as periodic tasks.

Prerequisites

Authorize DataWorks to access Platform for AI (PAI).

You can go to the authorization page to grant the permissions with a single click. For details about the permission policy, see AliyunServiceRoleForDataWorksEngine. Only an Alibaba Cloud account or a RAM user with the AliyunDataWorksFullAccess policy can grant these permissions.
A project directory has been created.
A PAI Designer node has been created. For more information, see Create a node for a scheduling pipeline.

Step 1: Configure the PAI Designer node

On the PAI Designer node editing page, configure the node as follows.

Node task

If you want to load an existing pipeline, you must first create it in PAI. You can then search for and load the pipeline by its name.

If the PAI Designer pipeline changes, click Reload to import the latest version. Then, click Publish to deploy the task. The page also provides shortcuts to Go to PAI Designer to edit and Go to PAI Designer to create.

On the PAI Designer node editing page, you can create a PAI Designer pipeline in one of the following ways:

Create a blank pipeline.

You can create a blank pipeline based on your business requirements and build a model by adding and dragging and dropping components. For more information, see Create a pipeline: Blank pipeline.
Create a pipeline from a preset template.

PAI Designer provides preset templates that you can use to quickly create similar pipelines. You can then modify the components or their configurations to build your model. For more information, see Create a pipeline: Preset template.
Create a pipeline from a custom template.

You can save a stable pipeline as a custom template for other members in your workspace to use and edit. For more information, see Create a pipeline: Custom template.

Scheduling parameters

If you need to periodically run a node task that you created, you can click Scheduling Settings on the right side of the node editing page to configure the scheduling information for the task based on your business needs. For more information, see Node scheduling configuration.

DataWorks provides Scheduling Parameters for dynamic variable assignment in periodic tasks. You can define variables in your task code using the ${var_name} format and then assign values to these variables in the Scheduling Settings > Parameter section. For more information about the supported formats, see Formats of scheduling parameters.

The following code provides an example of using a scheduling parameter:

--command='echo '\''${var_name}'\'';' \ -- Use with scheduling parameters.

After configuring the task, click Save.

Step 2: Publish and manage the node

After the node task is saved, you need to publish the node. For more information, see Node/Pipeline Publishing.
After publishing the task, go to the Publish to production environment section and click Go to operation and maintenance to view the task's status. For more information, see Introduction to Operation Center.

Note
You cannot run PAI Designer nodes directly from DataStudio. Instead, you must run them from the Operation Center. In the development environment's Operation Center, go to the Node O&M page and select Auto Triggered Task O&M > Auto Triggered Task. Then, find the task and click Test or Supplementary data in the Actions column to run it.