Re: [dnsext] New Version Notification for draft-ah-dnsext-rfc1995bis-ixfr-02

In message <4E525476.4070001@necom830.hpcl.titech.ac.jp>, Masataka Ohta writes:
> Shane Kerr wrote:
> 
> > Consider an environment where one has 2 distribution servers at a node,
> 
> Only 2 distribution servers? It should be a toy zone.
> 
> And, do you mean "zone", not "node"?
> 
> > If the node suffers a power failure, then it is quite possible that both
> > distribution servers will need a full AXFR when they restart.
> 
> Are you seriously saying that it is quite possible that two servers
> simultaneously loss checkpointing information because of power
> failure?

No.  I think Shane means that they have been out of contact long
enough that the AXFR style response is smaller than the IXFR style
response and the upstream supplied a AXFR style response.

> Even then, there is no point of IXFR-only because, with your example,
> neither server can offer IXFR.

You are missing Shanes point, which is that after the initial AXFR
style response that the slaves that transferred from the DM that
transfered first, and hence had a lower serial, will not be able
to get a delta style response from the other DM.

> > If zone
> > updates are frequent, there is a good chance that they will not have the
> > exact same starting serial.
> 
> So?

See above.

> > One solution is to have a single distribution server.
> 
> It's no solution. I'm afraid you completely misunderstand DNS.

Shane isn't saying that it is a desirable solution.

The other solution is to prevent the axfr's out from the DM's until they
both have the same serials available ixfr from.

The is not to say that I'm apposed to ixfr only.

> 							Masataka Ohta
> _______________________________________________
> dnsext mailing list
> dnsext@ietf.org
> https://www.ietf.org/mailman/listinfo/dnsext
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: marka@isc.org