Failure to acquire IP over DHCP

[Marko Vojinovic]

On Thursday, 30. August 2012. 11.43.46 Dave Ihnat wrote:
> Once, long ago--actually, on Thu, Aug 30, 2012 at 11:24:58AM CDT--Marko 
Vojinovic (vvmarko at gmail.com) said:
> > which dhcp servers are out there (DHCPDISCOVER), and asks anyone for
> > an IP assignment (DHCPREQUEST). Three servers respond: 137.138.16.6
> > and 137.138.17.6 refuse (DHCPNAK), while 192.168.0.1 accepts (DHCPACK)
> > and gives you an internal IP, which doesn't seem to be able to access
> > the outside world.  I am not sure if this is a misconfiguration or
> > implemented on purpose, ...
> 
> I would bet dollars to donuts they don't intend for that RFC1918 address to
> be out there.  It'd be extremely sloppy to have it on the same segment as
> the 137.138 addresses; it's more likely something they don't intend to run
> a DHCP server that's doing so anyway.

The first thing that comes to mind is that someone plugged in a small router 
there (unofficially, without proper authorization and configuration), between the 
"wall outlet" and the computer, and the router tries to be an "local" dhcp 
server. The MAC of the router is not in the CERN database, so the official dhcp 
servers refuse to cooperate, and neither the router nor the machine plugged 
into it can have access to the rest of the Internet. But that's just a wild 
guess... ;-)
 
> > The guys at CERN usually know what they are doing, ...
> 
> You'd certainly hope so!

Sure they do, I collaborated with them on a couple of occasions. Besides, they 
are one of the base developers of Scientific Linux (a RHEL clone used 
throughout CERN and related academia), so I certainly wouldn't consider them 
noobs or something... :-)
 
Best, :-)
Marko