What problem does this solve?

While using BroadwayRabbitMQ.Producerfor Broadway, we were not able to override the message acknowledger according to the docs as there is a unique :delivery_tag required to ack messages to RMQ. When we tried, this was empty and our messages were not being acked.

Second, the metrics were not showing up for the RabbitMQ backed pipeline, but they were for a pipeline using a normal GenStage producer and after some digging around, I noticed that the :name tag/label on the /metrics were different:

GenStage Producer Metric

broadway_process_message_duration_milliseconds_count{batcher="local",name="Prom.Pipelines.LocalPipeline"}

RabbitMQ Producer Metric

broadway_process_message_duration_milliseconds_count{batcher="rabbit",name="BroadwayRabbitMQ.Producer"}

Changes

After switching the tag_values function to use :topology_name, the metrics for both Pipelines started to show up accurately in Grafana.

Curious if there was a reason to use the message acknowledger for the name tags? BroadwayRabbitMQ emits different events for acking/ rejecting.

I'd be happy to add those in a separate PR. Wanted to keep this one short.

Issue number: N/A

Example usage

• I created an app with two pipelines, one with a GenStage Producer and one with a RabbitMQ Producer both with a 10% error rates.
• After making these changes, the metrics started to show up accurately in the Grafana dashboard, including exception rates and batch failures.

Additional details and screenshots

GenStage Pipeline

RabbitMQ Pipeline

Checklist

• I have added unit tests to cover my changes.
• I have added documentation to cover my changes.
• My changes have passed unit tests and have been tested E2E in an example project.