Re: [Rabbit-proxy-users] Empty files in cache

[Robert O. <d9...@na...>]

Hello!

> It was happening yesterday with IE, but now I can only seem
> to have a problem on Mozilla.

Ok, is this a page that you had in any web browser cache?

> Here's what HTTPSnoop saysa about the request from Mozilla.
> Mozilla gets a blank page:
>
> GET http://www.tco.net/ HTTP/1.1
> Host: www.tco.net
> User-Agent: Mozilla/5.0 (X11; U; NetBSD i386; en-US; rv:1.2.1) Gecko/20021221
> ...
> If-None-Match: "42801f-2313-3d1bf524"
> Cache-Control: max-age=0

Seems you have a cached copy and mozilla tries to revalidate the
cached copy. That should work I think.

And if you clear the mozilla cache everything gets ok once again.
Hmmm. Ok, this sounds like a bug, will have to investigate a bit
more. In the meantime dont mix surfing with and without proxy unless
you also clear the browser cache when you switch....

> Here's the request from Netscape which sees the page correctly:
This one does not have the page in its cache.

> And here is IE which also sees the page correctly:
Ok, also dont have a cache on disk...

> Here's Mozilla loading adios.tco.net which is the same file
> on the same server, just a different name.  This one displays
> perfectly:

Well, different hosts means different resources so they should be
treated independantly. You dont have that file cached in mozilla so
it does not try to revalidate.

> Interestingly, enough.  It says that http://www.tco.net/, it's
> linked stylesheet and all the images that it loads have null
> content type and 0 size.  However, the cache entries for
> adios.tco.net have the correct size and content type.

Given above info that makes sense, but should be marked as a BUG...

> Sometimes, it seems that the Status admin page keeps connections
> to www.tco.net listed longer than the browser appears to be
> connected, but I can't reproduce this reliably.

Well RabbIT uses keepalive whenever possible so that is as it should.
Keepalive to clients and keepalive to servers are two independent
things for a proxy. The keepalive time for servers is lower than
keepalive time for clients though.

> I imagine they were just small ones.  I didn't realize that it
> differentiated, but it makes sense.

Well, yes, its quite a heavy process to start an external program and
small images will not see much savings. Even if you save 50% its
still only 1000 bytes...

/robo