Mage Pro makes pipeline recovery, reprocessing, and historical backfills seamless. Instantly resume workflows from failure points, recreate datasets for any past period, or backfill years of data with precision and control.
Through built-in retry mechanics, partitioned backfilling, and powerful recovery options, Mage Pro ensures you can fix fast and recover faster—without re-running entire processes or losing progress.
Whether you’re debugging failures, reprocessing stale data, or migrating large datasets, Mage Pro empowers you to manage pipeline history and future execution effortlessly.
How it works
Automatic retry configuration:
Define global, pipeline-level, or block-specific retry settings using retry_config in your metadata. Automatically retry block runs on failure with customizable delay, retries, and backoff strategies.
Retry from failure points:
Instead of restarting entire pipeline runs, retry only the failed block runs or resume from a selected block within the individual pipeline run page.
Backfill pipelines easily:
Create backfills by:
Setting date and time windows (e.g., backfill daily data over a year)
Generating custom partitions through code (for dynamic backfill scenarios)
Custom runtime variables for dynamic backfills:
Access variables like interval_start_datetime
, interval_end_datetime
, and interval_seconds
inside Python or SQL blocks to build smart, time-aware backfills.
Structured retry logs:
Every retry is recorded in detailed logs, ensuring transparency and auditability.
Why it matters
Data workflows aren’t always linear—and when failures or gaps happen, restarting from scratch wastes time, compute, and budget. Mage Pro’s recovery and backfilling features offer:
Fast recovery without re-running successful steps
Flexible backfilling to restore missing or outdated data
Higher developer efficiency through surgical retries and partial re-execution
Lower operational risk by giving full control over history reproduction
Mage Pro helps you not just monitor history—but rebuild it better whenever you need.
Your AI data engineer