[ILUG] 302 redirection strangeness

[Paschal Nee]

Could this not just be a http proxy issue?  The proxies should not cache the
302 response (301 might be cached if that's what you want) but will cache
the response from serverB and (depending on your cache-control/expiry
settings) serve it up to other users using the same proxy without coming
back to serverB.

What kind of cache settings do you have?  Can you post the raw HTTP headers
being sent back from serverA and serverB for the same file?  Could the
errant IPs be corporate proxies?

Paschal.


On 28/03/06, ilug-request at linux.ie <ilug-request at linux.ie> wrote:
>
> >>> What is that image? Does it happen to be on other pages as well
> >>> so the
> >>> browsers might cache it and not download it again?
> >>
> >> Again, if the browser has cached the file, it won't request it, and
> >> hence won't get a 302 redirect.
> >
> > What if they are doing If-Modified-Since: GETs.  The first server
> > will redirect and the other will return 304 (I think) meaning not
> > modified, so you won't see it in the count of 200s.
>
> Ah - that makes sense, and my apologies to the previous poster who
> mentioned 304s as this is probably what he meant too, but I didn't
> understand him.
>
> However, sensible as it sounds, it's not what's happening :-(  My
> original post was incorrect (it was late :-( ) and the count of 200
> result codes I gave was in fact the count of ALL matches in the logs
> for the redirected files on serverB, which did contain a significant
> number of 304s. Apologies for the confusion, but the basic question
> remains the same - why do 25% of the requesters to the original
> server not follow the redirect?
>
>
>
> Niall
>
>
>
>
>
> ---------- Forwarded message ----------
> From: "Kenn Humborg" <kenn at bluetree.ie>
> To: "Niall O Broin" <niall at linux.ie>, "Irish Linux Users Group" <
> ilug at linux.ie>
> Date: Tue, 28 Mar 2006 15:45:58 +0800
> Subject: RE: [ILUG] 302 redirection strangeness
> > However, sensible as it sounds, it's not what's happening :-(  My
> > original post was incorrect (it was late :-( ) and the count of 200
> > result codes I gave was in fact the count of ALL matches in the logs
> > for the redirected files on serverB, which did contain a significant
> > number of 304s. Apologies for the confusion, but the basic question
> > remains the same - why do 25% of the requesters to the original
> > server not follow the redirect?
>
> Pull out the heavy guns...  Grab a tcpdump for a few minutes on the
> two servers and get down and dirty :-)
>
> If you can isolate a working and non-working case in the logs and
> look at the HTTP traffic for those requests in detail, you might
> find enlightenment.
>
> Later,
> Kenn
>