Mailing-List: contact dev-help@hc.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "HttpComponents Project" <dev@hc.apache.org>
MIME-Version: 1.0
From: Gary Gregory <garydgregory@gmail.com>
Date: Wed, 16 Aug 2017 15:00:50 -0600
Message-ID: <CACZkXPyXe4sdqMZs-_42Q7N=BdRR8FAqXhKcpJ5ausX+RBoUag@mail.gmail.com>
Subject: Core 4 and problem with socket timeout to origin server
To: HttpComponents Project <dev@hc.apache.org>
Content-Type: multipart/alternative; boundary="001a11441148d9528f0556e53061"
archived-at: Wed, 16 Aug 2017 21:01:02 -0000

--001a11441148d9528f0556e53061
Content-Type: text/plain; charset="UTF-8"

Hi All (Oleg, really)

I am wondering how to fix/deal with the a timeout occurring in an async
proxy when an origin server takes too long to reply.

You can reproduce this issue with our asyn example NHttpReverseProxy:

Start it with: http://postman-echo.com 8888

Then run:

curl --verbose --max-time 1000 http://localhost:8888/delay/3

and you will get an EMPTY response from the server, no HTTP status code
where I would expect a 504 or 500.

The proxy prints out:

Reverse proxy to http://postman-echo.com:80
[client->proxy] connection open 127.0.0.1:8888<->127.0.0.1:61128
[client->proxy] 00000001 GET /delay/3 HTTP/1.1
[client->proxy] 00000001 request completed
[proxy->origin] connection open 192.168.0.112:61130<->52.44.234.173:80
[proxy->origin] 00000001 GET /delay/3 HTTP/1.1
[proxy->origin] 00000001 request completed
[client->proxy] connection closed 127.0.0.1:8888<->127.0.0.1:61128
[client<-proxy] 00000001 java.net.SocketTimeoutException
[proxy->origin] connection released 192.168.0.112:61130<->52.44.234.173:80
[proxy->origin] [total kept alive: 0; total allocated: 0 of 100]
[proxy->origin] connection closed 192.168.0.112:61130<->52.44.234.173:80

Curl prints:

* STATE: INIT => CONNECT handle 0x600057930; line 1410 (connection #-5000)
* Added connection 0. The cache now contains 1 members
* STATE: CONNECT => WAITRESOLVE handle 0x600057930; line 1446 (connection
#0)
*   Trying ::1...
* TCP_NODELAY set
* STATE: WAITRESOLVE => WAITCONNECT handle 0x600057930; line 1527
(connection #0)
*   Trying 127.0.0.1...
* TCP_NODELAY set
* Connected to localhost (127.0.0.1) port 8888 (#0)
* STATE: WAITCONNECT => SENDPROTOCONNECT handle 0x600057930; line 1579
(connection #0)
* Marked for [keep alive]: HTTP default
* STATE: SENDPROTOCONNECT => DO handle 0x600057930; line 1597 (connection
#0)
> GET /delay/3 HTTP/1.1
> Host: localhost:8888
> User-Agent: curl/7.54.1
> Accept: */*
>
* STATE: DO => DO_DONE handle 0x600057930; line 1676 (connection #0)
* STATE: DO_DONE => WAITPERFORM handle 0x600057930; line 1801 (connection
#0)
* STATE: WAITPERFORM => PERFORM handle 0x600057930; line 1811 (connection
#0)
* STATE: PERFORM => DONE handle 0x600057930; line 1980 (connection #0)
* multi_done
* Empty reply from server
* Connection #0 to host localhost left intact
* Expire cleared
*curl: (52) Empty reply from server*

A bug in the example proxy? In HttpCore? Both? Or something missing in the
implementation of the example async proxy?

Thoughts?

Thank you,
Gary

--001a11441148d9528f0556e53061--