Submit an offline job

Overview

With this feature, you can use natural language commands to submit an offline job from the development environment to the O&M workspace. It supports common offline job types, including Shell, Python, Hive SQL, and MaxCompute SQL. During submission, you can configure the job's scheduling cycle (for example, daily, hourly, or minutely), time expression, upstream dependencies, runtime parameters, and priority. For minute-level scheduling, this feature supports two semantic modes: "continuous throughout the day" and "reset every hour". The agent selects the correct configuration based on your requirements.

Use cases:

• Submit a debugged offline job from the development environment to the O&M workspace for production scheduling.

• Configure a job's scheduling cycle, time expression, and upstream dependencies before submission.

• Submit a job with a specific minute-level scheduling configuration, such as continuous execution or hourly resets.

Prerequisites

• The target Dataphin project exists, and you are a member of the project.

• The file and directory for the offline job must exist in the development environment.

• Have the job code and necessary configuration details, such as the scheduling cycle, time expression, and upstream dependencies.

• For BASIC mode projects, you also need the job's node output name.

Examples

You can describe your requirements to the agent using natural language. The following are sample prompts that you can copy and use:

• "Submit this Shell job to the O&M workspace. Run it daily at 08:00 and 20:00."

• "Submit this MaxCompute SQL job, schedule it to run once a day, and set its upstream dependency to the project root node."

• "Configure and submit this job to run every 5 minutes continuously throughout the day."

• "Submit this Python job. Use the Python 3.11 engine and depend on the pandas library."

Interaction steps:

1. The agent typically confirms the target job, job type, code, and scheduling configuration. If any information is missing, it will prompt you for it.

2. Before submission, the agent verifies that the job file and directory exist. For minute-level scheduling, it confirms whether you want the "continuous throughout the day" or "reset every hour" mode.

3. The agent submits the job with the complete scheduling and upstream dependency configurations. This process replaces the entire configuration, not an incremental update.

4. After a successful submission, the agent confirms that a pending release record has been created and reminds you that the job requires a release operation to go live.

Parameters

This table lists the parameters for submitting an offline job. The agent automatically identifies the required information from your instructions and prompts you for anything missing. You do not need to provide system-generated parameters.