Why Flower makes its own connection to the broker instead of reusing Celery's connection?

[Roman513]

Hello!

Recently, I ran into an issue where Flower does not work well with an SSL Redis connection provided for Celery via a URL. Since Flower is a Celery app, Celery establishes the connection to the broker during initialization, and it successfully connects. I initially thought that Flower would use the same connection as Celery, but I discovered that it actually parses the Celery configuration and creates its own connection. However, for Redis SSL options, it only parses the broker_use_ssl config parameter - but not the Redis URL. This forces me to create a Python config for Celery, which is less convenient than simply passing the URL via environment variables.

I see two possible solutions:

1. Fix Flower’s broker config parsing to support all recently added Celery configuration features for broker URLs, including this one, in flower/utils/broker.py.
2. Remove this parsing and instead reuse Celery's existing broker connection or parser.

I assume there was a reason for creating this separate parser instead of just using Celery’s broker connection. Does anyone know the reasoning behind this and whether it's still relevant?

This might be a tough question, so perhaps @mher or others who have worked on this part, like @cyberjunk or @lpsinger, might have some insights.

Thank you!