Looks straight forward enough.
We used BFD MAD as opposed to ARP, tested it yesterday successfully with ZERO packet loss.
I went with BFD/MAD because I did not want to do the intermediate switch approach. Since the whole idea of using IRF was to get around having to use spanning tree, I chose not to do the ARP method as it required STP/MSTP to be configured.
It also appeared to me that if I used ARP/MAD that I would not be able to use both LACP links to the other switches. I do not know that definitively; but since spanning tree by its nature does not allow two links to be up simultaneously (and there was not alot of information to dispute this in the instance of ARP/MAD), I chose to go with BFD/MAD.
What was your thinking with going in that direction? Just curious...
The document for the A5800-A5800 posted here by pombeii is also something I used and is excellent. I liked the Virtual Connect and HP A-Series document because it provided me a mechanism by which to test the configuration which was an immediatley gratifying benefit to the customer.
Thanks,
P