Re: [squid-users] Web-polygraph testing

From: <>
Date: Wed, 14 Nov 2001 10:39:15 +1100


Your first guess was correct, I had accidently removed one of the content
options. It
just caught me out because the same workload was caching with another
proxy (BorderManager) but not with squid.

Any idea what would cause squid to vary largely in response rates?
Using a simple workload I was getting around 5000 responses a sec
initially, but running the same test the next day I was getting only
CPU / memory usage seems to be fine, and there is plenty of room in the
cache. There doesn't seem to be anything unusual in the logs either.


                    Alex Rousskov
                    <rousskov@measurement-fa To:
          > cc:
                                                    Subject: Re: [squid-users] Web-polygraph testing
                    13/11/2001 13:12

If client requests reach the cache, then the most likely causes are
           - not enough HTTP headers for Squid to cache
           - time de-synchronization between cache and servers
Are you using a standard workload like PolyMix? PolyMix configures
Polygraph to have enough headers for Squid to cache responses. Do you
use NTP to sync clocks? You should.

If the above does not help, enable Squid debugging to see why
responses are released from the cache.


On Tue, 13 Nov 2001 wrote:

> Hi,
> We are currently setting up Web-Polygraph to perform some proxy-cache
> testing. We have Web-prlygraph running on BSD and that seems fine. Our
> Squid is running on Solaris 2.6.
> Our problem is that when we run Web-polygraph with polysrv and then
> pointing to our Squid server, out squid server does not cache any
> The Web-polygraph test has each request set to 100%cacheability.
> What could be wrong?
Received on Tue Nov 13 2001 - 16:39:51 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:04:11 MST