Re: TTL/Auth Redux

[Pekka Savola <[e-mail]>]

Responding to both Daves..

On Mon, 14 Mar 2005, David Ward wrote:
> What you have written here Dave is what is stated in RFC 3682
>
> quoting:
>
> "3.  GTSM Procedure
>
>   GTSM SHOULD NOT be enabled by default. "
>
> So, it seems that the desire to change this RFC is being pushed through the 
> BFD singlehop.

I think the key thing to remember is why GSTM spec says that. 
AFAICS, they don't want to break backward compatibility with the 
existing base of BGP, MSDP, and whatnot protocols which didn't support 
GTSM from day one.

However, BFD did, so this should not be a technical factor.

Dave Katz:
> "SHOULD unless configured otherwise" isn't valid specsmanship.

The text was an oversimplification.  It could be said, for example,

                                                   All received BFD
    Control packets that are demultiplexed SHOULD be discarded
    if the received TTL or Hop Count is not equal to 255.
    If such packets are not discarded by default, the implementation
    MUST have a configuration toggle to enable the discarding.

or:

                                                   All received BFD
    Control packets that are demultiplexed SHOULD be discarded
    if the received TTL or Hop Count is not equal to 255.  If the
    implementation chooses not to do this by default, it MUST have a
    configuration toggle to enable the discarding.

(note, there appeared to be extra "the session" around 
'demultiplexed')

I could even live with the above wording if the first "SHOULD be 
discarded" was a MAY, i.e., if the still spec had a MUST for 
implementing the toggle for implementations which don't do it.

Otherwise, the text that dkatz proposed seemed pretty good.

-- 
Pekka Savola                 "You each name yourselves king, yet the
Netcore Oy                    kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings