Re: message to 'down' intf

[Fred Baker <[e-mail]>]

At 3:36 PM 3/27/95, Benny Rodrig wrote:
>In draft-ietf-rreq-cidr-02.txt, Revision 2.05 from 3/17/95
>Section 5.2.3, Local Delivery Decision, reads:
>              + The packet is delivered locally and not considered for
>                 forwarding in the following cases:
>                  - The packet's destination address exactly matches
>                    one of the router's IP addresses,
>
>Shouldn't this be unless the interface corresponding to this IP address
>is 'down'?
>If the router is configured with an IP address of net X, but is currently
>disconnected from network X, should it identify and deliver locally
>(e.g. answer ping) messages destined to that IP address?

Well, I don't think there's a place I can go to to say "yes" or "no" other
than this document, so perhaps it's your word against mine. But I know of
several routers that work by saying:

        receive the message
        figure out how to forward it
        it's going out interface X;
                is the destination address == "my" address on that interface?

which doesn't work when the interface is down.

I also know of an operational situation at my former company where the
customer got very irate when we had a bug in which if we were using no
routing protocol at all and an interface went down we would continue to
respond to pings on the down interface. We hadn't cleared the cache entry
in IP, because the routing protocol did that.

IMHO, no, one should not respond to  ping to a down interface.

>I also have an editorial comment:
>Section 5.3.5.4 mentions two types of broadcast that are indistinguishable
>under CIDR. This section is unclear, as the names of these two types
>were both changed to network-prefix-directed-broadcasts, making them
>truly indistinguishable...

argh; thanks for finding that.

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
computers run on smoke, it if leaks out they won't run