[squid-users] Workaround for failing hardware

From: Kurt Buff <kurt.buff_at_gmail.com>
Date: Wed, 25 Jun 2008 12:58:31 -0700

All,

I'm running squid 2.6.17 on a FreeBSD 6.2-STABLE. This should be
mostly irrelevant, as I am trying to ameliorate a hardware issue while
I get a new machine up and running. The issue is that we have
extremely slow initial load of pages, which I suspect is because of
slow and perhaps faulty disk access - haven't localized it to disk or
controller yet. We have moved from a T1 to a DS3, and ever since
people have been complaining of slow web access, which seems a bit
counterintuitive.

The hardware is:
     2.4GHz Celeron CPU
     2G RAM
     Single HD: ad6: 76319MB <WDC WD800JD-00MSA1 10.01E01> at
ata3-master SATA150

I'm getting a fair number of this message:

     ad6: TIMEOUT - WRITE_DMA retrying (1 retry left) LBA=112214751

and am noticing that squid is dumping core every once in a while,
though it does restart:
     pid 3509 (squid), uid 100: exited on signal 6 (core dumped)

I've tried to solve this temporarily by preventing squid from caching
at all - by adding "no_cache deny all" and "pipeline_prefetch on" to
the config.

What I want to make sure of is that other settings don't negate this
effort. What should I strip (or add) to my configuration below?

Current configuration file:

http_port 3128
hierarchy_stoplist cgi-bin ?
acl QUERY urlpath_regex cgi-bin \?
cache deny QUERY
acl apache rep_header Server ^Apache
broken_vary_encoding allow apache
cache_mem 512 MB
cache_dir ufs /usr/local/squid/cache 2000 16 256
logformat combined %>a %<A %ui %un [%tl] "%rm %ru HTTP/%rv" %Hs %<st
"%{Referer}>h" "%{User-Agent}>h" %Ss:%Sh
access_log /usr/local/squid/logs/access.log combined
access_log syslog combined
refresh_pattern ^ftp: 1440 20% 10080
refresh_pattern ^gopher: 1440 0% 1440
refresh_pattern . 0 20% 4320
acl all src 0.0.0.0/0.0.0.0
no_cache deny all
acl manager proto cache_object
acl localhost src 127.0.0.1/255.255.255.255
acl to_localhost dst 127.0.0.0/8
acl SSL_ports port 443
acl CONNECT method CONNECT
http_access allow manager localhost
http_access deny manager
http_access deny !Safe_ports
http_access deny CONNECT !SSL_ports
acl our_networks src 10.0.0.0/8 192.168.8.0/24 192.168.10.0/24
192.168.11.0/24 192.168.12.0/24 192.168.13.0/24 192.168.15.0
/24 192.168.24.0/24
http_access allow our_networks
http_access deny all
http_reply_access allow all
icp_access allow all
visible_hostname zsquid.mycompany.com
coredump_dir /usr/local/squid/cache
pipeline_prefetch on

Kurt
Received on Wed Jun 25 2008 - 19:58:42 MDT

This archive was generated by hypermail 2.2.0 : Thu Jun 26 2008 - 12:00:04 MDT