[squid-users] Very high CPU Utilization

Hi,

I am running squid-2.4.STABLE6 (RedHat's RPM) on RedHat 7.3 with stock
RedHat 2.4.18-18.7.x kernel.

The hardware configuration is as follows.
Compaq ML350 Server
Dual Xeon 2.2 GHz
2304 MB RAM
36 GB x 6 disks (4 used for cache + 1 for OS + 1 Spare)

The excerpt of squid configuration is as follows

cache_mem 128 MB
cache_dir diskd /var/spool/squid 30000 32 256 Q1=72 Q2=64
cache_dir diskd /var/spool/squid1 30000 32 256 Q1=72 Q2=64
cache_dir diskd /var/spool/squid2 30000 32 256 Q1=72 Q2=64
cache_dir diskd /var/spool/squid3 30000 32 256 Q1=72 Q2=64

Here is the output from cachemgr

HTTP Requests (All): 0.03622 0.04277
Cache Misses: 1.54242 1.54242
Cache Hits: 0.01745 0.01847
Near Hits: 1.54242 1.54242
Not-Modified Replies: 0.00463 0.00678
DNS Lookups: 0.00949 0.01153
ICP Queries: 0.00000 0.00000

Even during peak loads HTTP Requests (ALL) will not go more than 0.12 second
and Cache Hits will be hovering around 0.02XXX second

The cache comfortably does 170 req/second. But , at peak loads the CPU
utilization goes upto 99%.

What could be done to reduce the CPU utilization ?

Venkatesh K
Received on Thu Sep 11 2003 - 11:05:23 MDT