DNS usage in application protocols

[Greg Hudson <[e-mail]>]

Hi.  The following text appears in section 5 of
draft-ietf-drums-smtpupd-12.txt, regarding MX hostnames with multiple
A records:

	The destination host (perhaps taken from the preferred MX
	record) may be multihomed, in which case the domain name
	resolver will return a list of alternative IP addresses.  It
	is the responsibility of the domain name resolver interface to
	have ordered this list by decreasing preference if necessary,
	and SMTP MUST try them in the order presented.

My DNS background suggests that this "MUST try them in the order
presented" requirement is misguided; order of DNS records is not
supposed to be important.  The text has been copied (with s/MX/SRV/)
by a draft relevant to the IMPP group, so I'm concerned about a faulty
meme propagating.

Can people in this group comment on whether I'm off my rocker?  I
can't imagine a client wanting to tamper with the order of records
returned by a resolver, but if for some odd reason it does want to
(say, it happens to have some network map information, or a cached
indication of which A record worked last time), I don't see why it
should be a violation the standard.

Thanks.


to unsubscribe send a message to [e-mail] with
the word 'unsubscribe' in a single line as the message text body.