KEEP_ALIVE_TIMEOUT causing the changes in server (mode debug) to be delayed?

Describe the bug When running Sanic with the following options:

app.run(auto_reload=True, debug=True)

If I make any changes to my code and refresh, the new code is not taken into consideration. I need to wait a few seconds, and I’ll see the following message in the console:

[DEBUG] KeepAlive Timeout. Closing connection.

Reading the documentation, settings KEEP_ALIVE to False in dev mode would solve my issue, but it doesn’t change anything. The error still is displayed after about 8 seconds.

What can I do to have my new code refreshed immediately when running Sanic?

Expected behavior New code executed on refresh, not the previous one.

Environment (please complete the following information):

• OS: Fedora 24
• Version: 21.6.2
• Python: 3.8.10

[cnicodeme]

I can provide a reproducible example, but probably not until Monday, so don’t be surprised if I don’t post anything until then 😃

[Tronic]

Currently (with or without my changes) REQUEST_TIMEOUT is completely useless. It only affects someone typing in a request over netcat because normally all headers are received in a millisecond, and confusingly enough RESPONSE_TIMEOUT is the one that actually covers the request body. I might prefer a more simplified timeout system with the ability to implement advanced logic in a streaming handler or via a middleware, if needed (e.g. an app could reset the timer after every I/O if it wants). But obviously that is another topic and not something that I would look into right now.