Hi all, At our company we recently implemented an IBM MQ integration for Airflow and I would like to gauge interest in contributing this as a new provider package.
Motivation With the introduction of event-driven scheduling and the MessageQueueProvider abstraction in Airflow, it has become significantly easier to trigger DAGs from external message brokers (as described in Astronomer's guide on event-driven scheduling): https://www.astronomer.io/docs/learn/airflow-event-driven-scheduling Many enterprises still rely heavily on IBM MQ as their primary enterprise messaging backbone. However, at the moment there is no official Airflow provider supporting IBM MQ. Our implementation enables: * An IBMMQHook * A MessageQueueProvider implementation for IBM MQ * The ability to trigger DAGs from IBM MQ events * Standard producer/consumer patterns from within tasks This allows IBM MQ to function similarly to Kafka, SQS, etc., within the Airflow event-driven scheduling framework. Technical Details The implementation is built on top of the open-source IBM MQ Python wrapper: * IBM MQ Python (ibmmq) library: https://github.com/ibm-messaging/mq-mqi-python IBM has recently released and documented their modern Python binding here: https://community.ibm.com/community/user/blogs/dylan-goode/2025/10/16/new-python-binding-for-ibm-mq The hook supports: * Secure connections (TLS) * Queue get/put operations * Configurable polling behavior * Transaction handling where applicable The MessageQueueProvider implementation integrates with Airflow's event-driven scheduling so that DAGs can be triggered based on IBM MQ messages. Why this might make sense: * IBM MQ is still widely used in regulated industries (banking, insurance, government). * Many enterprises using Airflow also run IBM MQ. * This would allow IBM MQ to be a first-class citizen in Airflow's event-driven ecosystem. * The dependency is officially maintained by IBM and open source. I am willing to act as initial maintainer and code owner, of course this is purely a proposition. I would appreciate feedback on: * Whether there is interest in such a provider * If yes, whether it should live under apache-airflow-providers-ibm * And if we formalize this as an AIP or draft PR Happy to share a draft implementation through a PR if there is interest. Thanks! David
