On Sat, Feb 05, 2011 at 03:43:38PM +0100, Jan-Frode Myklebust wrote: > On Fri, Feb 4, 2011 at 9:37 PM, Willy Tarreau <[email protected]> wrote: > > Hi Jan-Frode, > > > > On Fri, Feb 04, 2011 at 03:15:49PM +0100, Jan-Frode Myklebust wrote: > >> We're trying to filter out HTTP HEAD requests using haproxy, > >> to make the proxy behave like shoutcast servers, ref: > >> > >> http://comments.gmane.org/gmane.comp.web.haproxy/2961 > > > > But why the hell are your trying to deliberately break HTTP ? > > Because that´s what shoutcast seems to be -- HTTP/0.9 with some custom > ICY-headers.
But what you showed were HTTP/1.0 requests. HTTP/1.0 and 0.9 are very different : - HTTP/0.9 has no HTTP version on the request line - HTTP/0.9 has no header, thus no empty line. If there is a version on the line, it's HTTP/1.x. Every subsequent line is a header line and the request only ends at the empty line. Haproxy supports HTTP/0.9 requests. So it will not wait for an empty line. You can try it if you want. However as most products, it upgrades them as 1.0 when forwarding them. > > It might be a lot better for the long term to fix the server > > instead, or you'll constantly have to resort to dirty tricks > > to achieve that. > > Unfortunately I don´t control the servers (or protocol). I have lots > of STBs on private networks (rc1918), no NAT, and need to give them > access to various shoutcast servers on the Internet. But I guess I´ll > need to find some other solution.. look into shoutcast proxy servers, > or see if we can configure the clients to use squid + the > upgrade_http09 option. > > http://www.squid-cache.org/Doc/config/upgrade_http0.9/ What's said there is interesting because apparently it's used to prevent upgrade on the response path, but it apparently still upgrades the request. > > From RFC2616 par.5.1.2: > > > > The methods GET and HEAD MUST be supported by all general-purpose > > servers. All other methods are OPTIONAL. > > Wrong RFC in this case.. :-) > > http://www.w3.org/Protocols/HTTP/AsImplemented.html But that's not the example you sent either ;-) See what setup you manage to get to work, especially if the servers accept HTTP/1.0. If so, maybe we can try to downgrade the response to HTTP/0.9 (which simply consists in stripping headers). Regards, Willy
