After creating a task flow, you can edit its configuration, trigger it manually using different run modes, and manage published versions—including checking modification status, comparing versions, rolling back, and exporting snapshots.
Edit a task flow
Only the owner of a task flow can edit it.
On the task flow editing page, click Go to O&M in the upper-right corner.
On the Task Flow Information page, click Edit.
In the lower part of the editing page, click a tab and modify the settings as needed.
Tab What you can configure Task Flow Information Name, owner, stakeholders, and scheduling settings. See Getting started, Configure timed scheduling, and Configure event scheduling. Variable Configuration Time variables and constants that are shared across all nodes. Reference variables in SQL statements using the ${name}format.Notification Configurations Success and failure notifications for the task flow, and timeout and alert notifications for individual task nodes or the entire task flow. Operations History of operations on the task flow, including the time, operator, and action recorded for each change. Execution Logs Execution logs for individual task nodes.
Run a task flow
Select a run mode from the upper part of the editing page. Each mode targets a specific scenario—choose the one that matches your situation.
| Mode | When to use it | Requirements |
|---|---|---|
| Try Run | You want to run the task flow immediately on demand. | None |
| Dry Run | Task Flow A depends on Task Flow B through a dependency-check node, but you need Task Flow A to proceed without actually executing Task Flow B. Dry-run Task Flow B to generate a scheduled trigger record, which lets Task Flow A run normally. | None |
| Run at a specific point in time | You need to reprocess data for a past or specific business date without modifying SQL statements or task configurations. | Define a time variable for the task flow and reference it in SQL statements. The variable value is calculated based on the day before the run date and the specified offset. |
| Run at a specific time range | You need to run the task flow across multiple consecutive business dates in sequence. Tasks run serially: each date must complete successfully before the next begins. | Define multiple time variables for the task flow. Limit: A maximum of 50 node instances can run in this mode. For example, if the scheduling cycle is one day, the task flow instance can run only for 50 days. |
Publish or unpublish a task flow
Manage versions
Check whether a task flow has unpublished changes
In the upper part of the editing page, hover over Publish. If the tooltip displays The task flow is modified. You must republish the task flow to make the modification take effect., the task flow has changes that are not yet published.
Compare the current version with the latest published version
When you modify a published task flow and click Publish, click Display differences in the Publish dialog box. The comparison view highlights all changes—a red dot marks each modified element.
View publishing records and manage published versions
Open the Published Tasks tab on the Task Flow Information page to see all published versions.
For each version, the following operations are available:
| Operation | Description |
|---|---|
| DAG | View the directed acyclic graph (DAG) of the task flow for that version. |
| Backfill Data | Generate one or more task flow instances for a specific point in time or time range, based on that version. |
| Rollback | Restore all task nodes and configurations to the selected version. After rolling back, republish the task flow for the change to take effect. |
| Export | Download the configuration file containing the edges and task nodes of that version. |
If a task flow has multiple published versions, you can select two versions and click Version Comparison to view the differences between the two versions.