prashant desai wrote:
> now the problem is after serving 500000 to 600000
> http requests squid becones very slow ,avg no of
> requests per day is aroung 100 to 150 thousend
Quite low load. That machine should not have any trouble to keep up.
What is your peak load?
In the snapshot from the cachemgr there was plenty of resources left. No
obvious sign of shortage.
Hmm. maybe you are running out of ports or some other socket resources.
What is the output of:
netstat -n | wc -l
netstat -n | grep SYN_R | wc -l
If the first is more than ~2000 then you need to tune
/proc/sys/net/ipv4/ip_local_port_range to allow the use of a larger
range of ports.
If the second is larger than ~80 (not very likely unless you have a lot
of high latency clients), then you need to tune up
/proc/sys/net/ipv4/tcp_max_syn_backlog.
-- Henrik Nordstrom Squid hackerReceived on Fri Feb 11 2000 - 12:16:52 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:51:11 MST