Hi,
I'm developing a proxy that it will work between Squid and another program. The problem is that with some sites how hotmail.com, google.com and others is calling, the proxy works very well, but with nba.com (p.ex) don't have answers.
The code is simple, I 'm working with multi-threading (I have a server and several clients) and I used select().
Why its happen? I'm expending lot of time to solve this problem and I don't know what more I need to search.
Thank you...
Hi,
When I test with the site nba.com and others that don't have reply I noted that the recv() = 0 (zero). Then, I would like to know when a socket in the read set closes the connection?
Thank you.
ok, after I understood!!!
I'm implementing it at this moment.
Thank you...
I am just curious if you are parsing the incoming request as it is arriving from the web browser... e.g. getting the target and starting the connect to squid prior to receiving the rnrn header & body seperator...
If so, try getting the entire request prior to establishing the outgoing connection to squid... then send it all at once and see if it makes a difference... again, the thinking is that the site is only allowing (or appears to) a second or less for the entire request and doing partial writes may be causing enough of a delay to overrun the server's allowed time... Of course, that would only happen if squid is doing the same thing i.e. if it grabs the target it immediately starts connection before it gets the full headers... and I haven't look at its source code for over a year so not certain...
freshmeat.net: Project details for TrustWall HTTP Proxy::
Apr 1, 2003 TrustWall HTTP Proxy protects your internal Web server by acting as an inbound proxy (like a reserve Squid proxy).
http://freshmeat.net/projects/twhttpd/HOME |
Forum Proxy Leecher Home - Overview (buy proxy,buy proxy list,buy ::
Forum Proxy Leecher Video Tutorial: http://download.my-proxy.com/ fpl_tutorial_zoom2.wmv. News: The 3d2f.com editors have called FPL the most promising in
http://fpl.my-proxy.com/HOME |
If not... still no clue... must think more... though it would be nice if they didn't have their server configured in such a messed up fashion... :-/
Michael
Hi,
I didn't understand your suggestion? Why it would work?
I did a couple of quick tests again the site... nba.com that is...
I can (programmatically) get the request out quickly enough to the system after establishing a connection... but that server appears to be set to a max 1 second wait before it terminates if it does not (at least start to) get the request headers...
For a stupid example... just telnet to the site (port 80) and watch how quickly it disconnects...
I do not have access to squid right now... but I would be curious as to the behaviour if you go from a web browser thru squid to nba.com ... if it works, that would mean there is a solution... if not, its a delay built into squid that is causing the problem... let me know and I will see if I can think of any fixes...
Charles : Web Debugging Proxy | HTTP Monitor | HTTP Proxy | HTTPS ::
Charles is an HTTP proxy / HTTP monitor / Reverse Proxy that enables a developer to view all of the HTTP traffic between their machine and the Internet.
http://www.charlesproxy.com/HOME |
Tinyproxy - Light-weight HTTP proxy daemon::
Tinyproxy is a light-weight HTTP proxy daemon for POSIX operating systems, released under the GNU GPL license.
http://www.banu.com/tinyproxy/HOME |
Michael
Hi!
Thank you for your attention! I solve the problem only working more with my threads. The problem was the threads synchronization.
bye
Be aware of the fact that nba.com does still appear to have apx. a 1 second timeout for the incoming request before it disconnects...
Either way... good luck with your project!
Michael
Hi, Michael!
I dont know if your answer will solve my problem because I don't change the headers in my request, I put it in a variable unsigned *char and I only send it to the Squid, the same occur with the Squid reply. But I will check it!!!
Thank you for the help.
Iliete
I am using a Web Browser thru Squid and I don't have problems... I only have problems when I put my 'proxy' between the Web Browser and the Squid... and I think how you, it must have a solution, it works with other sites!
Now I'm studying the Timeout, maybe I discover something...
Thank you more once...
I just caught the last messages in this thread...
The specs require that requests sent to proxy servers use an absolute uri... and since the impl in question is talking to a proxy (squid) the incoming uri shouldn't be modified...
A side note / fyi on this...
HTTP 1.1 requires that all in spec servers (proxy / origin) accept the absolute URI format in the request message... so if you know you're writing a non-chained proxy and know you're talking to a 1.1 system (e.g. via cached info from a prior request) then you don't have to do the re-write...
HTTP 1.0 does require that requests to non-proxy systems transform absolute URIs to an absolute path format with a Host: header field....
Michael
The most often seen reason for that type of behaviour is you are either not terminating the headers in your request with full rn sequences.... or that the responding server is not replying with them... the most common is to see termination being only n which is out of spec... but you have to compensate because MANY servers / clients are out of spec...
Without more info... thats about all I can say...
Michael
i think, your problem will be solved, when you change http://xxxx.com/sdfsd.fff to /sdfsd.fff or http://xxxx.com/ to /. thats all.
when you do a proxy server, you must implement right url change engine depend on rfc specification
 How much does getting a small tattoo on your hip/stomach hurt?
 Do anyone else have an itchy anus? ?
|
HTTP Proxy