Having exhausted all other possibilities I can test I have to ask:
Has anyone else ever done AoE through a bridge?
Because it does not seem to work. I have a feeling the AoE driver is
making assumptions about what kind of interfaces it will transmit and
listen on for AoE traffic.
I described my setup in an email yesterday.
And according to: http://wiki.xensource.com/xenwiki/XenNetworking
I should be able to create a bridge (I'll call it aoebr) and I should be
able to put eth1 (the interface plugged into the AoE SAN) and vif0.0 on
the bridge and talk to the bridge from veth1 which is the opposite end
of the vif0.0 "wire". I do not specify anything in
/dev/etherd/interfaces which means AoE should broadcast and listen on
When I do this and run tcpdump on the disk node (AoE target) I can see
the packets from the AoE initiator (Xen box with bridge) and it sends a
reply back. But the Xen box with the bridge never seems to see the reply
packet because it never recognizes the AoE disk.
If I put eth1 specifically in /dev/etherd/interfaces I get the same
Since I am supposed to be talking to the bridge via veth1 anyway I have
tried putting veth1 into /dev/etherd/interfaces but nothing arrives at
the disk node when I do this. In fact nothing seems to elicit any
network traffic at all except talking on eth1 and then only one-way
conversation. This makes me think that AoE does not work properly on
interfaces which are not actual physical ethernet cards.
I have made AoE work very successfully without bridging under a Xen
enabled kernel. So I don't think it is Xen itself. It seems that AoE
just does not like bridging. This is a big problem for me because my
whole project hinges on getting AoE to work over the bridge.
Can anyone confirm whether AoE can communicate over veth interfaces via
a bridge? ie. you have actually done it?
Tracy R Reed http://ultraviolet.org
A: Because we read from top to bottom, left to right
Q: Why should I start my reply below the quoted text