[squid-users] Problem with cache_peer

From: Gemmy <chenryn_at_163.com>
Date: Fri, 23 Jul 2010 00:49:03 +0800

 Hi~
I hava a squid cluster with two cache_peer using round-robin. Today, I
want to add the third nginx, but it's load average increases quickly,
and the netstat tell that the number of TIME_WAIT is more than 5000!
I run "squidclient mgr:server_list" on the squid servers.The "OPEN
CONNS" to old nginx is only digits, but three hundred to new nginx!
I collect and count the access.log too,the number of "PARENT/old" is
nearly same to "PARNET/new"...
The cache_peer configure like following:

cache_peer 10.2.3.4 parent 80 0 no-query round-robin originserver name=d1
cache_peer 10.2.3.5 parent 80 0 no-query round-robin originserver name=d2
cache_peer 10.2.3.6 parent 80 0 no-query round-robin originserver name=d3
cache_peer_domain d1 .test.com
cache_peer_domain d2 .test.com
cache_peer_domain d3 .test.com

And the nginx servers' configure are exactly the same.

Is there any wrong?
Received on Thu Jul 22 2010 - 16:49:09 MDT

This archive was generated by hypermail 2.2.0 : Thu Jul 22 2010 - 12:00:05 MDT