On Wed, 17 Nov 2010 18:03:45 +0100, Jakob Curdes <jc@xxxxxxxxxxxxxxx> wrote: > Hi, > > I am trying to work out squid's (3.0stable20) behavior when used for > (non-anonymous) FTP downloads. I use an own test ftp site for this, so I > can see the FTP logs. > > When I download an existing file, all is as expected. > > When I try to download a non-existing file, squid obviously tries to get > it from the origin server, which gives him the error code 550. > But this does not seem to be forwarded to the browser or other client; > the browser displays an empty page and when downloading with wget it says > "Proxy request sent, awaiting response... No data received." > Not even an HTTP header in return. > > I would have expected that squid passes on the error status or displays > its own error page as it does for http errors!? > We have a pretty standard config so I won't post it here for now. Any > ideas ? Or am I on the wrong track? > > Regards, Jakob Curdes There are bigger problems in 3.0. In short FTP in 3.0 and older require the username AND password to be publicly visible in the URL. Please try with 3.1 or 3.2. NP: Sadly there is one regression in the 3.1.8/3.1.9 packages which made it not challenge right, snapshot bundle is needed there for correctly working HTTP auth to an FTP server. Amos
