[j-nsp] M7i DHCP Relay

Chuck Anderson cra at WPI.EDU
Thu Aug 12 10:51:12 EDT 2010


On Thu, Aug 12, 2010 at 06:20:38AM -0700, Kaj Niemi wrote:
> On 12/8/2010 15:33, "Chuck Anderson" <cra at wpi.edu> wrote:
> 
> > 1. MX shouldn't require Option 82 (relay-agent-option) in order to
> > function as a stateless DHCP Relay Agent (BOOTP Helper), but it does.
> > 
> > 2. MX shouldn't get confused and fail to function when the edge switch
> > has added it's own DHCP Option 82 information to the packet.
> 
> On IRB, I *think* it must have relay-agent-option because it seems like it
> wants to write the interface name in the Option 82 packet no matter what and
> if the helper is stateless by itself it must somehow be able to figure out
> where (what interface) to return the packet to...

Yes, that is what it does.  However, my old Layer 3 switch had no such 
need to maintain the state in Option 82.  It just relayed via the 
GIADDR which is already in every BOOTP/DHCP packet... Why can't the MX 
lookup the GIADDR in the DHCP packet and send the response out the 
interface that is assigned the GIADDR's IP address?  Every other 
router or switch I've seen does this and has been doing this for 
decades.  BOOTP/DHCP isn't exactly a new protocol to have to know how 
to get right.

> With stateful dhcp relay that isn't an issue. I think it does count
> licensing wise (1 address = 1 subscriber license)?

Hopefully not, or Juniper may be supplying free subscriber licenses 
for us.  We shouldn't have to pay for a working DHCP Relay Agent.

If Juniper is serious about the Enterprise market, they need to fix 
some things.


More information about the juniper-nsp mailing list