Funky arp problem
long.kenny at gmail.com
Mon Mar 16 22:08:29 MDT 2009
Will you provide us with a bit more information? How many of these
misbehaving servers do you have? Are they all on the same broadcast or
switching domain or are they on different networks separated by routers? Do
they all start up the problem at the same exact time? Do you have an
unmanaged or managed switch that these are plugged into? When the problem
occurs, run a "tcpdump -i eth0 -vvvn" where eth0 is the NIC in question.
Then try to ping the server in question from another box on the network, and
see if the server in question is receiving the ARP requests. If its not, it
wouldnt seem likely the problem belongs to the server, but elsewhere in the
network. If you are seeing the ARP requests but your server is not
responding to them, then we can work from there.
I hope this helps you.
On Fri, Mar 13, 2009 at 5:20 PM, Mike Lovell <mike at dev-zero.net> wrote:
> Mike Lovell wrote:
>> I have several Supermicro servers. Over the past about 2 months, a few of
>> them have started being stupid when it comes to networking. On these
>> servers, I reinstalled the OS (debian etch and *gasp* windows) and,
>> afterwards, the box stops responding to arp requests. If I try to ping the
>> server, I get a no route to host error. The arp tables show an incomplete
>> for the ip address of the box. Funny thing is that if I then get to a
>> console for the box and just generate any kind of network traffic, things
>> then work. Any ideas on this? (and I don't really consider a cron to wget
>> google every minute a solution). It has happened on a few different models
>> of servers and across Linux and Windows. Thanks
> As a bit more information, it appears that all of the machines are running
> Intel 82563EB based network adapters.
> PLUG: http://plug.org, #utah on irc.freenode.net
> Unsubscribe: http://plug.org/mailman/options/plug
> Don't fear the penguin.
More information about the PLUG