Page Fault and configuration

From: Vincent Schultz <vincent.schultz@dont-contact.us>
Date: Mon, 19 Jul 1999 12:23:28 +0200

Hello Gentlemen,

I have just installed Squid 2.2 stable 2 on a Spark Station 10 with 128
Mo of RAM, one 75 Mhz cpu. There are 8 SCSI disks of 4Go (full at only
25%) on one controler.

The cache Manager ("General Runtime Information" menu) shows me some
results like :

Number of HTTP requests received: 209614
Page faults with physical i/o: 807958

The ratio seems *very* high as i read !

I read the FAQs and changed in my squid.conf the following things :

cache_mem 80 MB
memory_pools off
maximum_object_size 1048 KB

The top shows :

last pid: 17499; load averages: 0.22, 0.28,
0.35
11:57:47
87 processes: 86 sleeping, 1 on cpu
CPU states: 0.0% idle, 7.9% user, 10.4% kernel, 81.7% iowait, 0.0%
swap
Memory: 128M real, 1972K free, 217M swap in use, 526M swap free

  PID USERNAME THR PRI NICE SIZE RES STATE TIME CPU COMMAND
12518 squid 1 35 0 192M 87M sleep 121:31 11.83% squid

...and squid is a bit slow despite a 2 Mb leased line to our provider.

What command can help me more (sar -options?, vmstat) ?
Any idea or suggestion ?

Thank you for helping me,

Regards,

Vincent
Received on Mon Jul 19 1999 - 04:15:24 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:47:27 MST