Re: draft-jennings-app-dns-update-00

Cullen Jennings wrote:

> a HTTP based API to update DNS records similar to the system at
> dyndns.org.  Be pleased to get peoples thoughts

Interesting.  I've implemented a simple dyndns update client, the
main trick is to get this right in hardware, and then stick to it
for some *decades* (sic!).  Discuss it also with the dyndns folks.

In the record parameters I miss MX and SPF among others.  I'm not
sure about "MUST https", how about SHOULD ?  Dyndns allows more
than one "host".  You have that as domain, clearly you don't want
to limit it to hosts, but allowing more than one might be good.

In the examples ITYM one "?" to start the query part with the 1st
name=value pair, but later the conventional separator is "&", not
"?".  

Apart from 0 and -1, do those indices actually work with DNS ?
What a client sees as second record might not be the second
record in a RR set from a server's POV.

Example 4.2 actually defines two MX records, it doesn't "update"
them.  If the index is supposed to be either -1, 0, or n+1 (for
n existing records) this might work.  But then index=1 has the
same effect as index=0, and n+1 has the same effect as index=-1,
so something is still odd with those indices.

How about s/index/action/ and:
action=0 kill all, optionally create new first
action=1 keep all, add new record

If action=1 matches an existing record it's a noop, no harm done.

A very interesting part in the dyndns protocol is that clients
MUST NOT update anything unless it has changed.  They go as far 
as demanding to maintain persistent state (surviving a reboot).

They also allow to omit the IP, taking the IP of the query as
default (when update clients cannot figure out what the public
IP is).

 Frank

_______________________________________________
Apps-Discuss mailing list
Apps-Discuss@ietf.org
https://www.ietf.org/mailman/listinfo/apps-discuss