By default, configuration changes for scheduled DataWorks Tasks take effect in Next Day mode — Instances generated the day after deployment use the new configuration. If you need a new or modified Task to take effect on the same day, switch to Immediately After Deployment mode.
Choose a mode
| Next Day | Immediately After Deployment | |
|---|---|---|
| When changes take effect | Day T+1 (first Instance of the next day) | Day T (based on deployment time + 10-minute buffer) |
| Production stability | High — current-day Instances are unaffected | Lower — misuse can cause dependency conflicts or accidental Instance replacement |
| Recommended | Yes (system default) | Only for specific cases (see When to use Immediately After Deployment) |
How it works
After you modify a Task and click Submit, the selected mode determines when the new configuration takes effect.
Next Day (system default)
Deployment on Day T updates only the Task definition. All Instances already generated or scheduled for Day T continue to run based on the pre-deployment configuration.
Changes take effect starting from the first scheduled Instance on Day T+1.
Tip: To apply the new logic on Day T without switching modes, run a Backfill Data operation for the Day T Instances after deployment.
Immediately After Deployment
The system uses deployment time + 10 minutes as the cutoff to decide how to handle each Instance scheduled for Day T:
| Scheduled time vs. cutoff | Behavior |
|---|---|
| Later than deployment time + 10 min | Normal execution — the Instance is generated and runs based on the new configuration. |
| Earlier than or equal to deployment time + 10 min | New Tasks: The Instance performs a Dry Run (business logic does not execute). Its status shows as expired instance that is generated in real time. Modified Tasks: No new Instance is generated for that time slot. |
On Day T+1, all scheduled Instances are generated based on the new configuration regardless of which mode was used on Day T.
Limitations
| Limitation | Details |
|---|---|
| Daily cutoff window | The system runs batch Instance generation from 23:30 to 24:00 each day. Deployments submitted during this window take effect on Day T+2, regardless of which mode is selected. |
| Data Source changes | If you only modify the Data Source associated with a Task, already-generated Instances for the current day are not updated even with Immediately After Deployment selected. Those Instances continue to run using the previous Data Source. To apply the change immediately, run a Backfill Data operation. |
When to use Immediately After Deployment
Immediately After Deployment mode is higher-risk. If misused, it can create tangled Dependencies, accidentally replace or delete Instances, and destabilize the day's scheduled Tasks.
For Tasks that are already deployed and need modification, use the safer two-step approach: deploy with the default Next Day mode, then run a Backfill Data operation to manually trigger the Instances that need to run on the current day.
Use only for these cases
-
New Tasks with no complex Dependencies that must run on the same day as deployment.
-
Replacing existing Instances — swapping already-generated Instances for the current day that have not yet run.
Avoid in these cases
-
Modifying the Scheduling Configuration of a deployed Task — especially Tasks with complex Upstream and Downstream Dependencies. Changing the Scheduling Recurrence (for example, from daily to hourly) while deploying immediately can leave old Instances in place while generating new ones, creating dependency conflicts.
-
Inconsistent generation modes across dependent Tasks — if an Upstream Task uses
Next Daywhile a Downstream Task usesImmediately After Deployment, the Downstream Instance for the current day cannot find its Upstream Dependency and becomes an Isolated Task that does not run automatically.
Practical scenarios
Scenario 1: Deploy a new Task
After deploying a new Task, each Instance's behavior depends on how its scheduled time compares to the deployment time.
| Scheduling time | Status and behavior |
|---|---|
|
Later than (Deployment time + 10 minutes) |
The system generates a normal Scheduled Instance, which runs at its scheduled time. |
|
Earlier than or equal to (Deployment time + 10 minutes) |
The system generates an expired instance that is generated in real time. This Instance performs a Dry Run and does not execute. To process data for the current day, perform a Backfill Data operation for the previous business day. This operation also has a 10-minute time difference when generating the Instance. For more information, see How it works. |
Example: A Task deployed to the Production Environment at 12:00 has an effective cutoff of 12:10.
-
Instances scheduled after 12:10 run normally as Scheduled Instances.
-
Instances scheduled at or before 12:10 perform a Dry Run, with status expired instance that is generated in real time.
Scenario 2: Update scheduling recurrence
When you update the Scheduling Recurrence of a deployed Task and redeploy with Immediately After Deployment, old and new Instances may coexist on the same day. This only occurs on the day of deployment — Day T+1 generates Instances normally based on the new configuration.
-
Instances scheduled in the future: DataWorks replaces them with new Instances based on the latest Scheduling Configuration.
-
Instances already past their scheduled time: DataWorks keeps the Instances scheduled before the new time and replaces or deletes those scheduled after it.
Scenario 3: Inconsistent generation modes
If both an Upstream and a Downstream Task are new and use different modes — for example, the Upstream Task uses Next Day and the Downstream Task uses Immediately After Deployment — the Downstream Instance for the current day cannot find its Upstream Dependency. The Downstream Task becomes an Isolated Task and is not scheduled to run automatically.
If the Isolated Task has many Downstream Dependencies, it can prevent all subsequent Tasks from running.
Scenario 4: Change upstream scheduling time
When you modify the scheduling time of interdependent Tasks with different Scheduling Recurrences, Downstream Instances depend on both new and unreplaced old Upstream Instances, according to the latest Scheduling Configuration. This only occurs when Immediately After Deployment is selected and the scheduling time has changed.
For details on hourly and minutely dependency scenarios, see Must-read: Principles and examples for scheduling configuration in complex dependency scenarios.
Examples:
-
Upstream changes from every 6 hours to every 8 hours (00:00, 08:00, 16:00) with Immediately After Deployment selected.
-
Upstream changes from every 6 hours to once a day at 16:00 with Immediately After Deployment selected.
