Search squid archive

Re: Squid Accelertor Problem with Cache_Peer's

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Henrik Nordstrom wrote:

mån 2007-03-05 klockan 12:32 -0500 skrev Shane A. Froebel:

We at the battlestarwiki.org are using Squid in accelerator mode at the
moment to serve as the primary point for all web-connections. However,
everytime we have like a horde of users access the site, the server
itself goes slow. So I tried settings up a secondary cache server on our
SQL server (which only has SQL and no httpd). Now I read everything I
can and I can still not get everything to work correctly.

I don't quite get how you want the relations between the two.. should
users connect to one or both?

From your configs it looks like users always connect to the Squid on the
same server as the web server. Under these conditions it won't generally
help to have this Squid bounce the requests to yet another Squid.

What you can do to improve performance is to have another server also
accept requests from the users, and forward them to the web server.

But most importantly you need to diagnose why things gets slow. So I'll
follow up with a bit of questions focusing on your first server for now:

* Is there any swap activity under load?
* How is the Squid cache configured?
* What is the primary load condition? CPU? I/O? Memory?


Regards
Henrik
Apollo and Athena are two different machines.

1)
22:14:17 up 14 days, 28 min, 1 user, load average: 34.79, 28.25, 27.54 -- Been like this since after the last episode aired.
            total       used       free     shared    buffers     cached
Mem:       2067852     832508    1235344          0     104556     264192
-/+ buffers/cache:     463760    1604092
Swap:      2040212     355476    1684736
2) Right now Apollo uses the default dir /usr/local/squid/var/cache (5gb) and since we don't use much "data" space for files because of the mediawiki setup, a 10gb space under /home/squid_cache which is 10gb -- running AS root because it runs on port 80. Apache 1.3.3 runs on 80, but bind to 127.0.0.1 IP 3) Each machine has a Intel \ 2.4 Ghz \ P4 Celeron, and 2gb of ram in apollo and 1gb on athena. Both 100mb uplinks. Seagate \ 80GB:IDE:7200RPM Barracuda \ ST380011a - Redhat Enterprise Linux - OS ES.

Apollo has cPanel.

Talking with my friend, we are thinking about just changing to a more powerfull machine that can support muli-cpus. So any help would be great.

On athena, we have MySQL bind to one IP, and I have, in-thery, squid.battlestarwiki.org on another IP.

-- Shane


[Index of Archives]     [Linux Audio Users]     [Samba]     [Big List of Linux Books]     [Linux USB]     [Yosemite News]

  Powered by Linux