For AI agents: a documentation index is available at the root level at /llms.txt and /llms-full.txt. Append /llms.txt to any URL for a page-level index, or .md for the markdown version of any page.
SubDAGs are a legacy Airflow feature that allowed the creation of reusable task patterns in DAGs. SubDAGs caused performance and functional issues, and they were deprecated Airflow 2.0. Astronomer recommends that you don’t use SubDAGs and instead use an alternative supported Airflow feature.
In this guide, you’ll learn about SubDAG alternatives and SubDAG issues.
Assumed knowledge
To get the most out of this guide, you should have an understanding of:
Don’t use SubDAGs. Use one of the following alternatives instead:
Task Groups: Task Groups can be used to organize tasks in the DAG’s Graph View. Task Groups simplify the organization, viewing, and monitoring of complex DAGs. To view a video overview of Task Groups, see Grouping.
Cross-DAG dependencies: Cross-DAG dependencies can be implemented between different DAGs in the same Airflow environment or across separate environments. Cross-DAG dependencies are ideal if you have task dependencies that cannot be implemented within a single DAG. There are multiple methods available to implement cross-DAG dependencies.
SubDAG issues
SubDAGs DAGs are embedded within other DAGs and this can cause the following issues:
When a SubDAG is triggered, the SubDAG and child tasks occupy worker slots until the entire SubDAG is complete. This can delay other task processing and, depending on your number of worker slots, can lead to deadlocking.
SubDAGs have their own parameters, schedule, and enabled settings. When these are not consistent with their parent DAG, unexpected behavior can occur.
