POST doesn't close the connection when 204 response used

[billmote]

What you did:

Created a POST route
{ "id": "8e54c04a-21fd-4cb3-b8de-1aa5a9ef988e", "route": "/events", "httpMethod": "POST", "statusCode": "204", "delay": "0", "payload": { "success": true }, "headers": [ { "id": "6ffb195d-c434-46da-a08c-52090415c949", "header": "Retry-After", "value": "60" } ], "disabled": false }
that responds with a HTTP 204 with a simple body of { success: true } and posted:

curl --request POST '169.254.241.176:3000/events' --header 'Content-Type: application/json' --data-raw '[{"eventId":"fbf1de19-8629-45d3-ad0f-34b8e17d594f","sessionId":"07726e3d-68ed-49a2-bb37-d1372e941191","dateTime":"2020-08-14T01:45:44.975Z","eventType":"App Launched","category":"engagement","latitude":40.4543,"longitude":41.4563},{"eventId":"37314774-d52b-4219-b5c4-e3b2c14aeda0","sessionId":"48830e55-33b2-4a8e-b5c7-710b85b44bf5","dateTime":"2020-08-14T01:56:53.220Z","eventType":"view cart","category":"engagement","latitude":40.4543,"longitude":41.4563,"sku":"abc123","price":10.95},{"eventId":"6c8a2370-0fb0-4c47-8914-11a8e2ffe47f","sessionId":"48830e55-33b2-4a8e-b5c7-710b85b44bf5","dateTime":"2020-08-14T01:57:23.341Z","eventType":"App Launched","category","engagement","latitude":40.4543,"longitude":41.4563}]'

What happened:

The body isn't delivered for ~90 seconds -- my connection timeout on the client is set to 30 seconds.

Problem description:

Because it should immediately respond and close the connection?

Suggested solution:

Changing nothing other than the response code back to HTTP 200 fixes the problem (the route responds instantaneously), but I would like to emulate several of the available responses, and I specifically need the 204 response for this POST :)

[wohlben]

Hey @billmote, i'm not the maintainer but just stumbled upon this issue.

while I'd agree that a timeout is suboptimal in this situation, your desired behaviour is kinda confusing as well.

204 is explicitly for situations where you do not have a body.

There is no content to send for this request, but the headers may be useful. The user-agent may update its cached headers for this resource with the new ones.

you're likely suffering from apples decision how to treat these invalid requests.

Apple Safari rejects any such data. Google Chrome and Microsoft Edge discard up to four invalid bytes preceding a valid response. Firefox tolerates in excess of a kilobyte of invalid data preceding a valid response.

you should use another status code if you wish to respond with a body, for example 200 OK,

[billmote]

Interesting @wohlben. I will raise that with the people writing the endpoint. 204 shouldn't be an issue though (ref. https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/204). It just seems odd that it holds the connection open for a 204 and not a 200. In both cases the work is "done."

[wohlben]

eh, yes. thats exactly the documentation i linked earlier, and the source of my quotes as well.

for what its worth: i wasn't able to reproduce your timeout issue on my end. though mockIt never responded with a body either (despite having put that {"success": true} in there), only with the explizit status code, which is 204 No Content

$ curl -X POST localhost:3000 -vs
* Rebuilt URL to: localhost:3000/
*   Trying ::1...
* TCP_NODELAY set
* Connected to localhost (::1) port 3000 (#0)
> POST / HTTP/1.1
> Host: localhost:3000
> User-Agent: curl/7.60.0
> Accept: */*
>
< HTTP/1.1 204 No Content
< Access-Control-Allow-Origin: *
< x-api-key: testing123
< Content-Type: application/json; charset=utf-8
< Content-Length: 16
< ETag: W/"10-oV4hJxRVSENxc/wX8+mA4/Pe4tA"
< Date: Sun, 16 Aug 2020 19:17:28 GMT
< Connection: keep-alive
<
* Connection #0 to host localhost left intact

(same with your example request as well)

curl -vs --request POST 'localhost:3000' --header 'Content-Type: application/json' --data-raw '[{"eventId":"fbf1de19-8629-45d3-ad0f-34b8e17d594f","sessionId":"07726e3d-68ed-49a2-bb37-d1372e941191","dateTime":"2020-08-14T01:45:44.975Z","eventType":"App Launched","category":"engagement","latitude":40.4543,"longitude":41.4563},{"eventId":"37314774-d52b-4219-b5c4-e3b2c14aeda0","sessionId":"48830e55-33b2-4a8e-b5c7-710b85b44bf5","dateTime":"2020-08-14T01:56:53.220Z","eventType":"view cart","category":"engagement","latitude":40.4543,"longitude":41.4563,"sku":"abc123","price":10.95},{"eventId":"6c8a2370-0fb0-4c47-8914-11a8e2ffe47f","sessionId":"48830e55-33b2-4a8e-b5c7-710b85b44bf5","dateTime":"2020-08-14T01:57:23.341Z","eventType":"App Launched","category","engagement","latitude":40.4543,"longitude":41.4563}]'
* Rebuilt URL to: localhost:3000/
*   Trying ::1...
* TCP_NODELAY set
* Connected to localhost (::1) port 3000 (#0)
> POST / HTTP/1.1
> Host: localhost:3000
> User-Agent: curl/7.60.0
> Accept: */*
> Content-Type: application/json
> Content-Length: 717
>
} [717 bytes data]
* upload completely sent off: 717 out of 717 bytes
< HTTP/1.1 204 No Content
< Access-Control-Allow-Origin: *
< x-api-key: testing123
< Content-Type: application/json; charset=utf-8
< Content-Length: 16
< ETag: W/"10-oV4hJxRVSENxc/wX8+mA4/Pe4tA"
< Date: Sun, 16 Aug 2020 19:32:08 GMT
< Connection: keep-alive
<
* Connection #0 to host localhost left intact

and finally:

https://tools.ietf.org/html/rfc7231#section-6.3.5

A 204 response is terminated by the first empty line after the header fields because it cannot contain a message body.

/final edit:
there is an empty line transmitted, and the connection stays as keep-alive, so it is not according to specc, that seems to be true :)
though it is a pretty 'minor' thing

"Connection #0 to host (nil) left intact" means that although the transfer is over, the TCP session itself is still open (i.e no FIN/ACK exchanges have been made), allowing you to keep reusing the same TCP connection for multiple transfers (which could be useful if you don't want to sacrifice time on opening a new TCP connection).

[billmote]

I worked around the issue by adding a byteCount check before calling readAll(), which is what was getting hung up on the empty response. End of the day, makes my networking stack more resilient. Still talking through it with the service team, but here's my current config if you're interested:

{ "id": "8e54c04a-21fd-4cb3-b8de-1aa5a9ef988e", "route": "/events", "httpMethod": "POST", "statusCode": "204", "delay": "500", "payload": {}, "headers": [ { "id": "6ffb195d-c434-46da-a08c-52090415c949", "header": "Retry-After", "value": "5" } ], "disabled": false }

[boyney123]

@billmote wonder if it's worth adding this to the docs? What do you think?

Sounds like you got a work around, are you happy for me to close this issue?

[billmote]

I did work around it by not including a body IIRC. Anyone that has this issue will probably find this. Docs seem unnecessary.

…

On Tue, Nov 24, 2020 at 3:03 AM David Boyne ***@***.***> wrote: @billmote <https://github.com/billmote> wonder if it's worth adding this to the docs? What do you think? Sounds like you got a work around, are you happy for me to close this issue? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#80 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAMSPPANVGJORWADCSBGLCLSRNSEDANCNFSM4P7BHPBQ> .