[squid-users] squid process about 50% in kernel mode on solaris 5.8

Hi,

I'm currently having troubles with a Solaris 8 box that's running squid
version 2.4.STABLE2.
Sometimes the proxy was responding slow and when I started investigating I
found that the squid process itself was running something from 30-50+ % in
kernel mode.

A typical top output looks like this :

119 processes: 112 sleeping, 6 running, 1 on cpu
CPU states: 1.8% idle, 48.0% user, 50.2% kernel, 0.0% iowait, 0.0% swap
Memory: 512M real, 382M swap in use, 2086M swap free

  PID USERNAME THR PRI NICE SIZE RES STATE TIME CPU COMMAND
19296 root 1 11 0 322M 321M run 141:17 29.08% squid
 7886 root 1 0 0 1784K 1368K run 43:58 5.60% junkbuster
22946 beh_ops 1 50 0 2608K 1640K cpu 0:00 1.10% top
 7040 nobody 1 58 0 2344K 1896K sleep 1:24 0.31% dnsserver
  430 root 12 58 0 6248K 5384K sleep 31:16 0.23% mibiisa
22858 root 1 58 0 1792K 1008K sleep 0:00 0.20% junkbuster
  417 root 1 54 0 4120K 2664K sleep 74:06 0.15% opcle
 7041 nobody 1 58 0 2344K 1896K sleep 0:26 0.15% dnsserver
23135 root 1 0 0 1784K 1000K sleep 0:00 0.14% junkbuster
27139 root 1 58 0 1784K 984K sleep 0:00 0.06% junkbuster
11211 root 1 58 0 1784K 984K sleep 0:10 0.06% junkbuster
23134 root 1 0 0 1784K 984K run 0:00 0.05% junkbuster
23104 root 1 0 0 1784K 984K sleep 0:00 0.05% junkbuster
23132 root 1 0 0 1784K 984K sleep 0:00 0.05% junkbuster
19333 nobody 1 58 0 904K 600K sleep 0:09 0.03% unlinkd

It seems to me that squid is using a lot of memory.
I'm not sure if we're just too low on actual memory of if there are
configuration options I should set diffent.

Any hints are welcome.

Thanks,

mark

-- 
The From: and Reply-To: addresses are internal news2mail gateway addresses.
Reply to the list or to mbenschop <mark@cistron.nl>