Re: Overriding client reload

From: Kolics Bertold, University of Veszprem <>
Date: Sat, 2 Nov 1996 12:50:29 +0100 (MET)

Dear all,
> >> Corrupted in what way - "incomplete"? Isn't Squid only supposed to cache
> >> successfully downloaded objects?
> >
> >Thats the theory, yes... Unfortunately it doesnt always happen that way.
> >
> If you have (ideally reproduceable) examples of this, I'm sure Duane
> would like to hear about them. I so for presumed that objects from Squid
> are always intact, so the IMS scheme would make a lot of sense...
I have some examples:
the parent cache for domain is However guys
at helka use the latest cached from Harvest distribution (not the
commercial Harvest...). I sometimes notice that objects which is already
in the parent cache (FIRST_PARENT_HIT for us) are not the most update or
not the correct ones. If I reload with my browser, I get the same results.
However, if I connect to the local cache site and type the following
manually (to ensure that the specific object is reloaded on all level),
GET any_url HTTP/1.0
Pragma: no-cache

I receive the correct result (and of course any_url is accessed using the
parent cache).
I think, the bug is in harvest-cache rather than in the local squid-based

So, I think when using hierarchy of proxy caches, the softwares on each
level should be updated regularly.


 Kolics, Bertold E-Mail:
 University of Veszprem, Hungary W3:
 Information Engineering Course
Received on Sat Nov 02 1996 - 03:41:52 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:33:27 MST