Re: Set the IPv4 TTL to 1?
Robert Elz <[email protected]> Thu, 10 Feb 2005 04:30:40 +0700
| Newsgroups | gmane.ietf.zeroconf |
|---|---|
| Message-ID | <[email protected]> |
Date: Wed, 9 Feb 2005 10:41:09 -0800
From: Stuart Cheshire <[email protected]>
Message-ID: <[email protected]>
| The working group did consider whether "Link-local sources should specify
| TTL=1", and the proposal was rejected
| <http://www.drizzle.com/~aboba/ZEROCONF/ll29.html>. I'm not sure how it
| found its way into the document.
Yes, there was a big argument about this, which, if for no other reason
is justification for not changing anything this late in this area of the
doc.
| * Is this "advice" actually part of the formal specification or not?
| It doesn't say, "MUST", "SHOULD" or "MAY". Is an implementer required or
| expected to follow it, or not?
No, it is just a suggestion (or that's how I read it).
| * You say that application code should only set the TTL when "sending
| from an IPv4 Link-Local address"? If you've ever tried it, you'll know
| that on a multi-homed host, determining in advance what address you're
| going to be sending from is very difficult from user-level code.
Determining what will be picked for you is difficult. Determining
what should be used is trivial. If the application wants to send
from an LL address, and an LL address is available, forcing its use
is simple - as would be setting a TTL of 1 in that case.
That said, I doubt many applications will bother, I certainly don't
expect server applications to bother - but if an application writer
feels the need to consider which TTL should be set, when LL addresses are
in use, setting 1 is a reasonable choice (compat with followers of old
versions of the spec notwithstanding). [Aside: do you really expect
application writers to ever even look in this spec?]
| With all that said, if this is the only point of disagreement, then I
| will accept it and we should publish the RFC.
I think that is probably best.
kre
ps If/when this ever advances to DS, if implementations aren't documented
to be following this advice, then the proper procedure is to remove it.