Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 20 Nov 2002 06:10:40 +0300 (MSK)
From:      "."@babolo.ru
To:        Vincent Goupil <spoug@hotmail.com>
Cc:        freebsd-isp@FreeBSD.ORG, freebsd-net@FreeBSD.ORG
Subject:   Re: Slow network response with FreeBSD 4.6.2 and ipfilter
Message-ID:  <200211200310.gAK3AeSv006525@aaz.links.ru>
In-Reply-To: <F147ETM21UUflZnex440000ddb2@hotmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
> I have a system running FreeBSD 4.6.2-RELEASE-p5 #0 with ipfilter v3.4.27.  
> This system act as a firewall for an enterprise.  They need high 
> availability.  I have 5 network card, all 3C905 (3*3c905B-TX and 2*905C-TX). 
>   I made this setup in july and it run fine until 3 weeks ago.  The first 
> and second card are for the internet link (primary and backup).  The third 
> is for DMZ and the fourth is for local network.  The fifth is unused (marked 
> as down).  Each card as is own IRQ (except the fifth that is shared with the 
> first).  The high availability is provided by the two internet link, if one 
> goes down, the second take the load (change default route, ipf rules, ipnat 
> rules and DNS records).  This is done by a script running by cron.  We can 
> also do that manually.  We have two /29 network for the first link and one 
> /28 network for the second (we use alias on internet interfaces).  There is 
> only 3 services that run on the firewall: SSH (but only accessible from 3 
> subnets), ftpproxy (jftpgw 0.13.1) and snmp (only accessible by one subnet)
> 
> We begin to have problem 3 weeks ago.  The firewall begin to have a slow 
> response. I begin to have this arp message error (many times):
> 	arplookup 255.255.255.0 failed: host is not on local network
> 	arpresolve: can't allocate llinfo for 255.255.255.0rt
> We reboot the server and the network fast as earlier.  I finally find 
> something: when we use alias, we need to have at least one regular netmask 
> (instead of 255.255.255.255) for each network/subnetwork.  My error was on 
> the first link, my second sub-network was not configured properly.  I 
> changed it and it stop to have these errors about arp but the problem wasn't 
> resolved.  The network continue to be slow until we reboot the server.  This 
> happen during the day.  Now, it happen everytime.
> 
> What I've done:
> - I changed the netmask (as said earlier)
> - I upgraded from 4.6-RELEASE #0 to 4.6.2-RELEASE-p5 #0.
> - I look for IRQ conflict
> - I configure all interface with media and mediaopt.  They not using 
> autodetect anymore.
> - I chkrootkit and nothing found
> 
> What I suspect:
> - I read in a forum that the driver (xl) of 3C905 is not the best for 
> FreeBSD.  I don't know if this apply to 4.6.2.
> - Ethernet cables (I need to change it)
> - We run SSL (with a lot of users) in one of our web servers in the dmz. As 
> I know, SSL run on top of TCP, it should not be a problem.
> - When i run ifpromisc (in chkrootkit), it tell me that "xl0 is not promisc" 
> and "xl1 is not promisc".  I have 5 interfaces, what about the others ?
> 
> Can someone have an idea ?
What you mean when say "Slow network response"?
If that mean that packets trawel long
from some host to host under question
as reported by tcpdump, does ifconfig xlN down
and then ifconfig xlN up repare situation
for some time?
What tcpdump -npi xlN ether broadcast and not ip
say when slowdown hapens?

-- 
@BABOLO      http://links.ru/

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200211200310.gAK3AeSv006525>