Behaviour of UDP_HIT_OBJ requests ?

From: Andrew Cormack <scoanc@dont-contact.us>
Date: Thu, 15 Aug 1996 09:14:44 +0100 (BST)

I've noticed an unexpected behaviour in my squid 1.0.5 cache's handling
of UDP_HIT_OBJ replies, and would welcome confirmation whether it is
intended or accidental.

If my cache gets a large document from a neigbour (through UDP_HIT
followed by a GET) then the document is not cached by my cache, and the
next browser request for the document will have exactly the same result.
If on the other hand the document is small enough to fit in the UDP reply
packet (ie the response is UDP_HIT_OBJ) then my cache *does* save it, and
can handle future requests for the document without reference to the
neigbour.

This behaviour is a logical extension of the UDP_HIT_OBJ scheme to
minimise traffic in small objects (at very little cost in local disk
space) but it does lead to several neigbour caches having copies of the
same document, whereas for larger documents only one would keep a copy.

Before I include this behaviour in a document, could someone confirm
whether it is likely to stick around in future versions of the software ?

Many thanks,
Andrew

--
       +-----------------------------------------------------------------+
       | Andrew Cormack               | email: cormack@cardiff.ac.uk     |
       | Information Systems Support, | snail: 40/41 Park Place, Cardiff |
       |    postmaster and webmaster  |        CF1 3BB                   |
       | University of Wales, Cardiff |                                  |
       +-----------------------------------------------------------------+
Received on Thu Aug 15 1996 - 01:16:29 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:32:48 MST