[squid-users] Re: problem with cache box over 100req/sec

Happy New Year everyone!

I have setup my new customise machine to run squid.

I have been running quid over 2 year now.

For the first time, I was able to stress test my box to see how much request
the squid box can handle.

My machine is an IBM Netfinity 3000 (400MHz PII, 768MB Memory, 2 NIC cards,
2 15GB Ultra100 drive partition 4 time reiserfs.)

I have setup bridging and the diverter on my system running 2.4.20

Using Polygraph default simple.pg file with req=0.4 and robots=150+, I loses
connection to the box. In other words, my network functionality on my box
stops.

When it does, restarting the network does not bring it up.

I also notice that active connections remain in the system until I restart
the system.

I have tune my system which took me from 80req/sec to 120 semi stable box.

How can I stop this problem since it is not squid.

I have follow all tweaks offer by squid and also squid running on linux.

I will send my squid.conf and my sysctl file tomorrow.

Thank you very much.

Best regards

Edward Millington
BSc, Network+, I-Net+, CIW Associate
Systems Administrator, Sr
Cariaccess Communications Ltd.
Palm Plaza
Wildey
St. Michael
Barbados

Phone: 1 246 430 7435
Mobile: 1 246 234 6278
Fax: 1 246 431 0170

edward@cariaccess.com
www.cariaccess.com
Received on Thu Jan 23 2003 - 03:01:37 MST