I am getting this problem and I have just confirmed that our upstream ISP uses a transparent proxy. Is there a way to force a transparent proxy to not cache the data?
cheers, Iain. On Wed, 27 Nov 2002 02:39, Chris Hastie wrote: > On Tue, 26 Nov 2002, "Myles Pope (MLPA)" <[EMAIL PROTECTED]> wrote > > >This md5 sum 'bug/issue' may be related to the fact that some ISPs have > >a 'transparent' proxy inline downstream. (Mine does) > >Perhaps the md5 calc is coming off an old cached version? > >I believe that if you append a "?" on a url that it bypasses the proxy > >and so I suspect it may be possible to hack the freshclam code to do > >something like this for freshclam retreivals. > >Does anyone agree this is worthwhile and know how to go about it? > > I have had precisely this problem in the past, but since the > introduction of the --disable-cache option, which is now the default > since 0.54, things have been fine here. > > The problem manifested in a couple of ways. 1 - freshclam refused to > update, saying everything was already up to date. This occurred because > it checks the md5 from the server, and was getting a cached version. 2 - > It recognised the need to update, grabbed the 'new' database and then > complained that it didn't match the md5. This occurred when the md5 it > got from the proxy was new but the database was still old. > > The changes made to freshclam since should have sorted this out, but > transparent proxies are tricky beasts. I've also suggested ensuring the > server sends suitable cache-control and expires headers would help, > which is easily arranged on an Apache server by adding > > | ExpiresActive On > | ExpiresDefault A0 > > to the .htaccess file in the appropriate directory. > > When the problem was bugging me I found I could often get somewhere by > viewing the database and md5 files in a browser and hitting ctl+refresh > to force the proxy to refresh - though since I believe that's exactly > what the --disable-cache option did it may be won't work for you. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
