I’d appreciate if you guys could document your platform limitations a little better.
There were no response to this post, but we found out the hard way that there’s a 200kb request body limit, and it seems like a TCP reset will happen when exceeded.
That was causing broken pipe errors on our golang http client.
It is not documented anywhere so I end up spent like half a day to debug, nail down the issue, test and verify it.
It’ll save us and our customers a lot of back and forths + time if this was simply documented.
That’s not a limit. Something else must be going on.
Request bodies are streaming, there’s a little bit of buffering going on, but that’s only for fly-replay header users.
Are clients uploading these bodies particularly slow?
Can you tell us which app this is happening on?
I see you’ve linked to a repo which I can only assume is a reproduction, but can you also outline how you’re interacting with the flydebug app to recreate the issue?
Mind if I try this in a couple days time?
Hopefully before the bot closes the post.
Just got other urgent matters to take care of before revisiting this issue.