For context:
I started using n8n back in mid 2021, first with self hosted instance for personal use, then introduced it at work too. I consider myself as quite technical I work in the IT field as well and I am an experienced self hosted. I read documentations before asking, I use various LLM agent too to help me, and yet …. after few weeks of trying and trying again I gave up on 2.x and rolled back to 1.x
I am not claiming that I know it all, but I think that this point we can’t just call it “user error” the self hosting docs is not great.
Now I work with event driven architectures so this whole setup of broker, worker, runner, queue etc is not unheard of, however up till this point my setup worked just fine with a single container, my workload does not even justify database so I just used sqlite3. So apart from the connected services, ( redis for kv store instead of variables, gotify for notification, etc ) my setup was super simple, As a Python user I already used python code nodes already with commands even before having it in n8n even as just Beta, but when I tried to move to 2.x I just could not make it work.
So … saying all that, I would be more then happy to get support and a working example of 2.x with docker compose and blind mounts and python nodes, I would be even more happy if the whole AI workflows would not steal all the love and attention , and leave a little bit for what used to the core of n8n, the core I do love.
Keep it up !