airflow-ha

Monitor health with Apache Airflow and branch to stop or retrigger a DAG.

HighAvailabilityOperator runs a Python check returning (Result, Action). Health and recovery are separate: a failed result can retrigger the DAG while its current run follows a failure branch. Time limits stop monitoring, and maxretrigger bounds recovery attempts while preserving the health result.

Documentation

• Tutorial: run a DAG that retriggers itself builds a countdown with three runs.

• How-to guides cover service monitoring, alerts, Python and airflow-config, and stop boundaries.

• API reference defines results, branches, limits, callbacks, task models, and run configuration.

• Why health results and recovery actions are separate explains retries, retriggers, and workload ownership.

Published documentation is available at airflow-laminar.github.io/airflow-ha.

Related integrations

airflow-supervisor and airflow-nomad use this sensor inside external workload lifecycles and add log forwarding and diagnostics. airflow-pydantic provides declarative task models; airflow-config loads those models from YAML.

License

Apache 2.0. See LICENSE.