[squid-users] Finally I got it works

From: Davide Vaghetti <davide@dont-contact.us>
Date: Wed, 21 Jan 2004 09:56:33 +0100

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I had a big performance problem with my squid. Previously I was running
it on RedHat9 AMD athlon 1800+ some 18 Gbyte of cache and 256 Mbytes of
TOTAL ram.

The squid daemon starts responding to requests very slow after 1 hour I
had started it --- indipendentely from the number of users !. Every
time. I read a loto of guides/tutorial, I checked for bugged
ram/disks/whatever, but I can't solve the problem. I switched to
async-io and aufs, but nothing.

I started thinking the cache was too big (too much Gbytes to index), so
I reduced the size, but nothing.

Finally I worked on assigned RAM. In my default configuration I assigned
64 Mbytes of RAM to cache_mem, thinking that 1/4th of the total RAM was
a conservative approach. I was wrong !
It is true that a big cache dir require more RAM, but also the system
itself requires a lot of RAM, and all guides and tutorial I have read
are too optimistic regarding the RAM squid should use.

I finally reduced the cache dir to 3 Gbytes and RAM to 32 Mbytes ---
Squid never slow down and is from 10 to 100 times faster.

Now squid performances are directly related to the number of users.

- --
Davide Vaghetti
University of Pisa NOC - Centro Serra
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFADj7BxKJAsKiy+1ARAoTmAJ9xiz+/lKP5ktxIeWjnhd7Uq6dl0ACfTC/I
m8/TCOe5yph8bwA1tI6IeWs=
=0Tbz
-----END PGP SIGNATURE-----
Received on Wed Jan 21 2004 - 04:16:45 MST

This archive was generated by hypermail pre-2.1.9 : Sun Feb 01 2004 - 12:00:07 MST