Fwd: [squid-users] Re: wildcard purges /esi notes

i've asked to be removed countless times. here's another message that
i didn't want. it's really not that hard to remove somebody is it...

This is a forwarded message
From: Gabriel Wicke <groups@gabrielwicke.de>
To: squid-users@squid-cache.org
Date: Wednesday, January 28, 2004, 6:38:11 AM
Subject: [squid-users] Re: wildcard purges /esi notes

===8<==============Original message text===============
On Tue, 27 Jan 2004 23:47:49 +0100, Henrik Nordstrom wrote:

> See the purge script (Related Software)
>
> Not very efficient. Would be more efficient to monitor store.log and keep
> a database on what URLs there is in the cache (and in which objects).

Thanks for the pointer. I have no C++ experience so far, this might be an
opportunity to learn some (at some stage ;-).

Some notes on ESI (based on playing with it for an hour): if you have
trouble to get your ESI-enabled content cached, make shure your
Surrogate-Control header looks like this:

Surrogate-Control: max-age=60+60, content="ESI/1.0"

The '60+60' looks strange, but it's not at all (see
http://www.esi.org/architecture_spec_1-0.html#s2-2-3). Performance drops a
bit with ESI: about 600/s without ESI vs. 130/s with four unconditional
includes on my small laptop. Note that the ESI number is without
compression, 600/s would be hard to reach without.

-- 
Gabriel Wicke
 
===8<===========End of original message text===========
-- 
Best regards,
 mortbox                            mailto:mortbox@gamebox.net