Hello
I recently noticed a weird problem I was experiencing in my xen setup.
I'm actually not sure if the problem was with Xen, or with linux
bridging, but I want to let everyone here know, in case others have
the same problem.
My dom0 is running Xen3.0.1 on Ubuntu 5.10, with two physical nics.
One nic, eth0 is for management, ssh access to dom0, nfs traffic, etc.
The other nic, eth1, is for all domU traffic. It is in promisc.
mode, passing all traffic to the bridge, xenbr0.
The problem I was experiencing was that some domUs would not be able
to get through to the outside. I ran tcpdump on eth1, xenbr0, and
vif1.1 (or whatever the appropriate vif was for the domain in
question). The packets would appear on vif1.1, but not xenbr0 or
eth1. However, some domUs would work fine, and their packets would
show up on all three appropriate places.
I traced the problem to the MAC addresses. I need to specify mac
addresses for overall management on the greater network, but it
doesn't matter (to me or the network admin) what they are, as long as
they are unique on the network. It turns out that addresses starting
with 11:13 - 11:17 don't work. (There are probably other ranges that
don't work, but that is all I tested.) According to
http://standards.ieee.org/regauth/oui/oui.txt those addresses are not
assigned (though they are also not reserved for private use, or for
Xen), and I know that they are not in use on my network. I changed
the MAC addresses for the hosts that didn't work to a range that I
know works, and everything is fine.
Does anyone know why this happened?
In any case, I hope this info can save some other people a few headaches.
-Patrick
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users
|
Home