How do I set log levels for API middleware without flooding production?

[Lily.D]

I’ve been trying to implement a proper logging system for my API middleware, but I’m stuck on how to structure the log levels effectively. My debug logs are cluttering production, but I’m worried I’ll filter out something important if I just silence them.

[CharlesNL]

I’ve been wrestling with log levels in prod. Debug feels like noise, but I worry about missing something if I just silence it.

[Gary.T]

We turned off default debug in prod and added a per request trace flag behind a feature toggle; after a week the metrics looked steady and nothing obvious dropped, which was oddly reassuring.

[LukeUL]

Is the real issue maybe not the level but what we actually log? Could be the wrong fields or missing a correlation id.

[RyanFR]

I started drifting off topic for a day and then came back to the basics—tagging requests with an id helped sort through storms even when the logs got lean.