RE: [squid-users] huge page faults

The machine it runs on is a Celeron 533 with 256Mb Ram, 20Gb IDE, 3Com
ethercard
OS is Linux Redhat v6.2
Squid (2.5DEV-PRE4) is configured with 64MB memory / 4 GB cache

Ifconfig shows :
eth0 Link encap:Ethernet HWaddr 00:01:02:0C:81:9C
          inet addr:138.249.128.14 Bcast:138.249.128.255
Mask:255.255.255.0
          UP BROADCAST RUNNING PROMISC MULTICAST MTU:1500 Metric:1
          RX packets:4267417 errors:0 dropped:0 overruns:0 frame:0
          TX packets:3408483 errors:0 dropped:0 overruns:0 carrier:942
          collisions:67635 txqueuelen:100
          Interrupt:5 Base address:0xe400

regards,
./koen

Van Bossche Koen wrote:
>
> Hi all,
>
> I have a proxy configuration with huge page faults. from cachemgr :
> Number of HTTP requests received: 182337
> Number of ICP messages received: 390939
> Number of ICP messages sent: 393396
> Number of queued ICP replies: 23
> Request failure ratio: 0.00%
> Page faults with physical i/o: 20051330
>
> I believe Page faults means that memory is checked for an object, but at
> that time the object already is swapped to disk, right?

 You leave much to guess, such as squid version, os , os -version,
 machine ,machine - type, disk ,memory; all of these playing a role
 in this discussion.

 The 'simple page faults' are not necessarily related to squid's
 algortyhmic problems or issues.
 Just to low phys. mem can cause this, it should not necessarily be
relatd
 to high level application issues.

>
> What could from your experience be the cause of such high values.
> I also have a lot of collisions on my eth0, could this be related?

 Not probably,depends on which network the machine is.
 I don't think that a squid box could generate collisions if it would
 be alone on a network.

 M.

>
> regards,
> ./koen

-- 
 'Time is a consequence of Matter thus
 General Relativity is a direct consequence of QM
 (M.E. Mar 2002)