RE: Configuration tips for Squid 1.1.15

From: Larmour, Jonathan <[email protected]>
Date: Thu, 13 Nov 1997 20:51:18 -0000

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

You could tweak the build to ensure the compiler has maximum
optimisation on, e.g. with -O3. You could also use -fomit-frame-
pointer which would help, but which makes debugging impossible.
However this means you may as well compile without -g and can strip
the binary afterwards, although I wouldn't think this would make much
difference, if any.

There may also be other switches for gcc that are good for Alpha's.
You'll just have to look at the gcc man page.

Find out if there are any tunable parameters for your kernel and tune
them! There should be a lot of them. Presumably you are using RAID,
in which case you may want to look in the squid archives for
discussions about the optimal stripe size.

Why are there so many local domains? This will slow squid down. If
they really are local, since you are using ICMP then you should be
going to the source in those cases anyway.

Obviously there are also many things that can be tweaked in
squid.conf, but you'd have to post it here first (without the
comments please).

Oh, and upgrade to 1.1.17 :-).

Jonathan L.
Origin, 323 Cambridge Science Park,Cambridge,UK. Tel:+44 (1223)
423355
 ---[ It is impossible to enjoy idling thoroughly unless one has ]---
 ------------[ plenty of work to do - Jerome K. Jerome ]-------------
Fight spam! http://spam.abuse.net/ These opinions are all my fault

- ----------
From: Steve Ollis
Sent: 06 November 1997 23:58
To: 'squid-users@nlanr.net'
Subject: Configuration tips for Squid 1.1.15

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

Hi there..

I've been playing with Squid for a little while now (from 1.1.0
onwards), but have only recently
had the pleasure of having a dedicated platform to install it on.

I have a 433Mhz Alpha with 256 Mb memory, and around 13Gb of ultra-
wide SCSI storage on a
separate F/W SCSI controller.

I've compiled Squid with GCC and dmalloc, with the following Makefile

defines
HOST_OPT = # -DCACHEMGR_HOSTNAME="getfullhostname()"
AUTH_OPT = # -DUSE_PROXY_AUTH=1
LOG_HDRS_OPT = # -DLOG_FULL_HEADERS=1
ICMP_OPT = -DUSE_ICMP=1
DELAY_HACK = # -DDELAY_HACK=1
USERAGENT_OPT = # -DUSE_USERAGENT_LOG=1
KILL_PARENT_OPT = # -DKILL_PARENT_OPT
USE_POLL_OPT = # -DUSE_POLL
# do NOT define both USE_SPLAY_TREE and USE_BIN_TREE at the same
time!
USE_SPLAY_TREE = -DUSE_SPLAY_TREE
USE_BIN_TREE = # -DUSE_BIN_TREE
RELOAD_INTO_IMS = -DRELOAD_INTO_IMS

I want to know: Is there a better configuration to make the cache
perform better?
There are about 150-200 domains in the local_domain configuration,
and we have a reasonable amount
of traffic per day (ie recent peak of up to 543000 hits/day)

+-----------------------------------+-------------------------------+
| Steve Ollis | Email: ollis@att.net.au |
| Team Leader, Systems Engineering | Pager: steve@pager.att.net.au |
| AT&T EasyLink Services, Australia | Phone: +61 2 9911 1606 (BH) |
| Level 8, 15 Orion Rd | +61 2 9911 1555 (FX) |
| LANE COVE, NSW, 2066, Australia | +61 417 269 202 (MB) |
| "It's All Within Your Reach..." | #include "std_disclaimer.h" |
+-----------------------------------+------------------------
- -----BEGIN PGP SIGNATURE-----
Version: PGP for Personal Privacy 5.0
Charset: noconv

iQA/AwUBNGHNIk2HA9VDgtnSEQLhsgCgqVIGkvMu3DZyZqcglS7C9XpDZAwAnisP
vWWzbsBUznBx/joPmA4BJXE3
=nSWE
- ---
-----BEGIN PGP SIGNATURE-----
Version: PGP for Personal Privacy 5.0
Charset: noconv

iQA/AwUBNGtnkYYLUv2rigzBEQILNACbB5q1GYa9s2VPtiOdUwA7JN0CPeoAn0L2
QML3LEQpsFQDEik7v8dOLrb+
=i5so
-----END PGP SIGNATURE-----
Received on Thu Nov 13 1997 - 12:59:52 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:37:31 MST