Linux Netfilter discussions
 help / color / mirror / Atom feed
From: Vlad Adomnicai <vlada@xana.ro>
To: Netfilter Mailing List <netfilter@lists.netfilter.org>
Subject: Re: High CPU usage + Kernel option
Date: Tue, 06 Apr 2004 17:59:45 +0300	[thread overview]
Message-ID: <4072C5E1.60306@xana.ro> (raw)
In-Reply-To: <1081262572.24338.37.camel@raylinux.internal>

With vmstat the sys is at 100%. User process is 1-2%.
The machine doesn't do anything except routing and filtering with iptables.
There is also a script that runs every minute that updates the iptables 
rules, but it only lasts for about 1 second under medium cpu load.

Vlad Adomnicai


Ray Leach wrote:

>On Tue, 2004-04-06 at 15:35, Vlad Adomnicai wrote:
>  
>
>>Hi,
>>   I have a K6/2 333 machine with 64Mb of RAM and two network cards. 
>>(3c509 and an Intel one both with TCP cheksum offloading and Cpu )
>>   I use Fedora Core 1 with the default kernel    and iptables 1.2.9.
>>
>>   At high traffic through the router (6-7Mbytes/second) the CPU goes to 
>>100% and I can't even log on to it through SSH:
>>[root@root web]# ssh 192.168.200.1 -C -v
>>OpenSSH_3.5p1, SSH protocols 1.5/2.0, OpenSSL 0x0090701f
>>debug1: Reading configuration data /etc/ssh/ssh_config
>>debug1: Applying options for *
>>debug1: Rhosts Authentication disabled, originating port will not be 
>>trusted.
>>debug1: ssh_connect: needpriv 0
>>debug1: Connecting to 192.168.200.1 [192.168.200.1] port 22.
>>debug1: Connection established.
>>debug1: identity file /root/.ssh/identity type -1
>>debug1: identity file /root/.ssh/id_rsa type -1
>>debug1: identity file /root/.ssh/id_dsa type -1
>>    and stands there until a timeout occures.
>>   On the network behind the router are aproxymately 200 users for which 
>>I have about 200 iptables rules  like this iptables -A FORWARD -s <ip> 
>>-m mac --mac-source <mac> -j ACCEPT and 200   iptables -A FORWARD -d 
>><ip> -j ACCEPT, to allow passage only for the machines with the corect 
>>pair of ip/mac. I could give up the last 200 rules, as they don't serve 
>>a real purpose in limiting the access but they are used only for 
>>bandwidth monitoring / ip.
>>   Does anyone know how to lower the cpu usage with this configuration? 
>>    
>>
>It should be very low ...
>
>  
>
>>tweaks of any kind? Would a 2.6 kernel improve the situation? I have 
>>also seen an option in the 2.4 kernels CONFIG_NET_HW_FLOWCONTROL 
>>(Forwarding between high speed interfaces) but there it is written that 
>>it supports only some network devices and I don't know about 3coms or 
>>intel ones.
>>
>>  Any one has any ideas? another way of setting the rules? another 
>>filtering method? tweaking parameters? or at least what kind of system 
>>will it be necessary for this setup to be able to at least log on to the 
>>machine and do something on it. Also, would a FreeBSD be more suitable 
>>for this on the same configuration?
>>
>>    
>>
>Run something like sar, vmstat, top on the machine during high usage to
>see if there is another proc running that may be causing the high cpu
>usage. Do you run squid on that machine? If so, check the memory config
>...
>
>  
>
>>Thanks in advance for any informations.
>>Vlad Adomnicai
>>
>>    
>>




  reply	other threads:[~2004-04-06 14:59 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-04-06 13:35 High CPU usage + Kernel option Vlad Adomnicai
2004-04-06 14:42 ` Ray Leach
2004-04-06 14:59   ` Vlad Adomnicai [this message]
2004-04-08 19:24     ` danyvip (at) pattco.ro

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4072C5E1.60306@xana.ro \
    --to=vlada@xana.ro \
    --cc=netfilter@lists.netfilter.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox