channels_rabbitmq

channels_rabbitmq copied to clipboard

Hi @adamhooper 👋

We're about to release the first version of the Redis PubSub layer. I'm hoping that'll work out well, and it will become the default recommendation. But then, I'd also like to doc the RabbitMQ layer here as a More Serious™ option (if folks need that).

Would you be up for adding it to the Channels docs, at least to the community projects section, but arguably in the main layers pages too. (What it is and why you'd want it.)

Also, related, the PyPI page doesn't link back here, which would be handy I think.

Thanks

Hi @carltongibson,

Thanks for the attention! I'd love to be added.

I examined the Redis PubSub layer code. It's much cleaner than the old one. You don't have group_expiry any more, which is a huge win. However, some users may run into problems because there are no client-side constraints like capacity, channel_capacity and expiry. One slow HTTP client can eat all the server's RAM by reading messages too slowly.

With that in mind, here's my blurb:

channels_rabbitmq is a RabbitMQ-based channel layer focused on reliability. Use it to keep your site online and gain peace of mind. It does not support Task Queues: instead, it coexists with Celery.

Is there anything more I can do to help?

Hey @adamhooper — super — I'll add this to the docs this week (closing this issue in a PR) Thanks!