On Fri, Dec 24, 2004 at 05:41:45PM -0800, John Klos wrote:
> Hi,
> 
> I have an amd64 system running NetBSD 2.0. So far, I've found two bugs, 
> one of which may or may not be amd64 specific.
> 
> The first bug is that the first stage bootloader waits until something 
> happens with the keyboard before it proceeds. All of my colocated machines 
> are non-x86 hardware, but most people go for the lowest common 
> denominator, so I would guess that it could be a problem to have 
> production machines which need a human and a keyboard present in order to 
> reboot.

This may be specific to your hardware ...

> 
> The second problem is that I had let my amd64 machine get a DHCP lease, 
> then later gave it a different static IP on the same subnet. I started 
> using it as an NFS server, and had three machines communicating fine with 
> it. When I tried a fourth, I noticed that that machine could not 
> communicate via IPv4 with the amd64 machine. I checked and double checked 
> everything, but the only clue I could find was from errors in demsg:
> 
> arplookup: unable to enter address for 67.101.178.23@00:00:c5:43:3b:48 on 
> sip0 (host is not on local network)
> arplookup: unable to enter address for 67.101.178.23@00:00:c5:43:3b:48 on 
> sip0 (host is not on local network)
> arplookup: unable to enter address for 67.101.178.23@00:00:c5:43:3b:48 on 
> sip0 (host is not on local network)
> 
> The machines could communicate via IPv6 just fine. I checked the netmasks, 
> checked that both machines could communicate with all of the other 
> machines, then finally realized that the amd64 machine might not be able 
> to communicate with 67.101.178.23 because it used to have that address. 
> Even though I did an ifconfig sip0 delete before reassigning an IP, the 
> system still had some residual connection with that IP.
> 
> I rebooted, and now the two machines can communicate via IPv4 and IPv6 
> just fine. This could be a bug with sip*, or with amd64 - not sure. I've 
> not seen this on any other architecture or ethernet interface.

It could be that there was still a route to localhost for this address.
A 'route delete' would probably have fixed it.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--