[squid-users] analyzing cache in and out files

[Leonardo Rodrigues]

Em 30/09/15 04:13, Matus UHLAR - fantomas escreveu:
>
> the problem was iirc in caching partial objects
> http://wiki.squid-cache.org/Features/PartialResponsesCaching
>
> that problem could be avoided with properly setting range_offset_limit
> http://www.squid-cache.org/Doc/config/range_offset_limit/
> but that also means that whole files instead of just their parts are
> fetched.
>
> it's quite possible that microsoft changed the windows updates to be 
> smaller
> files, but I don't know anything about this, so I wonder if you really do
> cache windows updates, and how does the caching work related to 
> informations
> above...

     yes, i'm definitely caching windows update files !!

[root at firewall ~]# cd /var/squid/
[root at firewall squid]# for i in `find . -type f`; do strings $i | head 
-3 | grep "http://"; done  | grep windowsupdate | wc -l
824

     and yes, i had to configure range_offset_limit:

range_offset_limit 500 MB updates
minimum_object_size 500 KB
maximum_object_size 500 MB
quick_abort_min -1

(being 'updates' the ACL with the URLs to be cached, basically 
windowsupdate and avast definition updates - the second one required 
further tweaks with storeid_rewrite for the CDN URLs)

     from access.log, i see a lot of TCP_HIT/206 (and just a few 
TCP_HIT/200), so it seems squid is able to get the fully cached file and 
provide the smaller pieces requested:

[root at firewall squid]# grep "TCP_HIT/" access.log | grep windowsupdate | 
wc -l
9860
[root at firewall squid]# bzcat access.log.20150927.bz2 | grep "TCP_HIT/" | 
grep windowsupdate | wc -l
38584

     having squid to download the WHOLE file at the very first request 
(even a partial request) may be bad, but considering it will be used 
later to provide the data for other requests, even partial ones, make 
things a little better.

(this windowsupdate caching is running just for a few weeks, i expect 
HITs to grow a little more)


-- 


	Atenciosamente / Sincerily,
	Leonardo Rodrigues
	Solutti Tecnologia
	http://www.solutti.com.br

	Minha armadilha de SPAM, NÃO mandem email
	gertrudes at solutti.com.br
	My SPAMTRAP, do not email it