Backfill jobs
Overview
Backfill Jobs is a feature in the AE DataOps Platform for handling missing or updated data in specific scenarios. When data for a historical time range is incompletely calculated or produces incorrect results because of a system failure, a program exception, a business logic change, or another reason, a backfill job can reprocess and batch backfill the data for that time range.
By configuring a backfill strategy, you can precisely trace back data pipelines and efficiently rerun Flows, so that the data results for the specified time range are consistent with the latest logic. This ensures the completeness, accuracy, and consistency of data in your data warehouse or data lake, and provides key support for the continuity and reliability of business analysis.
Create a backfill job
On the Ops → Backfill Jobs page, click + Create Backfill Job in the upper-right corner to open the creation page
Backfill job settings
| Field | Description |
|---|---|
| Backfill Job Name | Unique within the space. Can contain lowercase letters, numbers, and underscores, and can't exceed 80 characters |
| Related Flow | You can select an "Online" Flow as the "source Flow" for the backfill job |
| Backfill Method |
|
| |
| |
| |
| View DAG | The system automatically parses and highlights the DAG based on "Select Node", so you can view the DAG relationships of the Flow |
Backfill job strategy
In Backfill Job Strategy, you can select the range of backfill base dates and the run rules of the job.
| Field | Description |
|---|---|
| Base date ${bd}: The base date for data settlement, and the unit for selecting backfill dates | |
Backfill Time Range |
|
| Date Setting Method |
|
| Order |
|
Run in Parallel | Default: On (Yes)
|
| Exception handling strategy | If backfill job instance fails
|
Backfill job statuses
| Backfill job status | Description | Available operations |
|---|---|---|
To be submitted | Draft status. Not yet submitted for execution, with no run instances |
|
| Running | After the job is submitted to run, its status becomes "Running" |
|
| Finished | The backfill job ran to completion successfully |
|
Failed | If the Flow instance for any base date ${bd} fails to run, the backfill job is considered failed |
|
| Terminating | Transitional status while the job is being terminated | - |
| Terminated | If you click "Terminate" while the job is running, the backfill job is terminated |
|
Backfill job management
On the Ops → Backfill Jobs page, you can manage all backfill jobs in the space. On the list page, you can view how backfill jobs ran, rerun jobs, manually terminate running jobs, and more.
The management fields are as follows:
| Field | Description |
|---|---|
| Backfill Job ID | Auto-increment unique ID |
| Backfill Job Name | Backfill Job Name |
| Backfill Method | Supported: Whole flow, Node only, To node, From node |
| Status | Includes 6 statuses: To be submitted, Finished, Running, Terminating, Terminated, Failed |
| Time Cost | Total run time of all latest Flow instances under the job (including queuing time) |
| Owner | Creator of the backfill job |
| Related Flow | The "source Flow" associated with the current backfill job |
| Created at | Creation time of the backfill job |
| Start Time | Time when the latest Flow instance under the backfill job started running |
| End Time | Time when the last Flow instance under the backfill job finished running |
Last Modified Time | Time of the last edit and save
|
| Action | Edit, run, delete, rerun, and more |
| Progress | Execution progress of the backfill job |

