Force-rerun descendant instances triggers a rerun of all downstream instances of an auto triggered instance across one or more workspaces. When you initiate the operation, DataWorks sets all targeted descendant instances to Not running and forcibly stops any that are currently running.
Prerequisites
Before you begin, make sure that:
Your DataWorks service is Professional Edition or a more advanced edition. If not, upgrade to Standard Edition or a more advanced edition first. See DataWorks edition selection and billing.
Your account meets one of the following role requirements: a RAM user with a workspace administrator or tenant administrator role, or an Alibaba Cloud account.
The auto triggered instance you want to rerun from meets both of these conditions:
Data timestamp is the previous day or the day before the previous day
State is Succeeded or Failed
Limitations
When filtering by workspace, you can select only workspaces where you are an administrator.
Each tenant in a region can run this operation only once at the same point in time.
If a descendant task has Rerun set to Disallow Regardless of Running Status in the Schedule section of the Properties tab, the force-rerun fails for instances generated by that task. To fix this, open the task's Properties tab, go to the Schedule section, and change the Rerun setting. See Rerun instructions.
Force-rerun descendant instances
Open the Force Heavy Run Downstream dialog box. In Operation Center, go to the Cycle Instance page and find the auto triggered instance whose descendant instances you want to rerun. Open its directed acyclic graph (DAG), right-click the instance, and choose Emergency Operations > Force Heavy Run Downstream. If the Force Heavy Run Downstream option is dimmed, the instance does not meet the eligibility requirements. Check that the data timestamp is the previous day or the day before the previous day, and that the instance state is Succeeded or Failed.
Configure the Basic Is Set step.
Parameter Description Data timestamp The data timestamp of the auto triggered instance. Only instances with a timestamp of the previous day or the day before the previous day are eligible. Start instance The instance from which the force-rerun starts in the current workspace. Rerun start point Whether the rerun starts from the current auto triggered instance or from its level-1 descendant instances. Rerun cross-day instance Whether to rerun Instance B (current-day data timestamp) when it depends on Instance A (previous-day data timestamp) and you rerun Instance A. 
For example, if Instance B is a daily-scheduled task that depends on Instance A, setting this to Yes reruns Instance B after Instance A completes.
Click Next, then click Confirm in the Please note message.
ImportantAfter you confirm, the settings in the Basic Is Set step are locked and cannot be modified. Verify your settings before confirming to reduce unnecessary computing and wait time.
Select workspaces to filter which descendant instances to rerun. In the Workspaces section on the left, select the workspaces whose descendant instances you want to rerun, then click
to move them to the Selected Workspaces section on the right. Descendant instances in selected workspaces are force-rerun. Descendant instances in unselected workspaces are dry-run only.You can select only workspaces where you are an administrator.
Click Next, then click Confirm in the Please note message.
Review the Confirm Influence step.
Parameter Description Total forcefully rerun instances The total number of instances to be force-rerun. Affected baselines Baselines whose output data will be affected by the rerun. See Overview. Force rerun downstream reason The reason for force-rerunning the descendant instances. 
Select the confirmation check box and click Confirm Rerun. DataWorks sets the targeted descendant instances to Not running and begins the force-rerun.
Example
Scenario: Force-rerun the descendant instances of an auto triggered instance generated by Node A, but only for Workspace A.
Result: The instance generated by Node C in Workspace A is force-rerun. The instance generated by Node B in Workspace B is not rerun — it is dry-run only.
What's next
View the details of this operation on the Operation History page. See View Operation Center operation records.