Hi Manav,

>
>I believe we are missing two nexthops that are quite extensively used.
>
>1. nexthop for which the associated ARP is in the stale state. These are
>special nexthops in the sense that in addition to forwarding traffic we
>must punt a copy to CPU so that the ARP can be refreshed.

NB> The RIB model deals with Layer-3 and above information (for the most
part). Handling of L2 state machine is beyond the scope of this document
(IMO)Š.just like handling of OSPF state machine is outside the scope of
this document. If a ARP entry becomes unresolved, then the nexthop becomes
unresolved and the i2rs agent needs to notify the controller of next hop
being unresolved. Internally the network device needs to do whatever it
does to refresh the ARP entry. This special nexthop is part of the
internal FSM on the network device.

Note that we are not trying to build a controller that controls each and
every thing on the network device. If we were building that, then what you
are suggesting needs to be done.



>
>2. nexthops associated IP interfaces created over vlans or VPLS services
>(extensively used). In this case the local route needs to point to a set
>of logical or virtual interfaces.

NB> A tunnel identifier needs to be addedŠ.I had it an earlier revision of
the draft and removed it by mistake.



>
>3. In Sec 2.4.5 the second bullet says:
>
>"DISCARD_WITH_ERROR: This indicates that the network device should drop
>the packet, increment a drop counter and send back an appropriate error
>message (like ICMP error)."
>
>Whats the ICMP error message that is being referred to here?

NB> Examples include ICMP_REDIRECT or ICMP Destination Unreachable. Exact
specifics needs to go in data-model.


Thanks
Nitin



_______________________________________________
i2rs mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/i2rs

Reply via email to