I have a couple FX5620's that have Realtek 8139 nics. I know that those
nics have problems and are not the best, but I am invested in the
FX5620's an cannot just throw them away. And I also know that it isn't
freebsd's or pfsense's fault that those nics have a problem.
I'm running embedded pfsense (1.0.1 and todays snapshot 2/14/07) clean
install with no configuration changes.
I have noticed that I can lock up the machines withing 5 seconds to 5
minutes by running a ping flood from the firewall to a laptop attached
to the lan (rl0) port with a cross over cable.
ping -s 30000 -f 192.168.1.254
The machines lock up hard, no errors on the serial console or on the
video console. From the windows task manager, (networking) the
fastethernet connection is 90% utilized when the lockup happens. I
wonder if the lockup has something to do with fragmentation since i'm
trying to send out 30k icmp packets and the (mtu is 1500), which get
split up and it is just too much to handle? Could all the packet
reassembly be the problem? Generating too many interrupts?
When I turn on polling the lockup problem disappears and the utilization
goes down to 25% no matter how much data I try to push. Does that mean
that it is an interrupt problem? This fixes the problem as far as I am
concerned since the most traffic these boxes will need to deal with is
5Mbps.
Is this a known and beat to death issue?
Thanks
Josh
--
Lake Agassiz Regional Library - Moorhead MN larl.org
Josh Stompro | Office 218.233.3757 EXT-139
LARL Network Administrator | Cell 218.790.2110
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]