> But, as with any IPv6 implementation: what needs to be tested first is to
> see whether stuff breaks when the answer for pool.ntp.org could have an
> AAAA record. What happens when the answer for pool.ntp.org contains one or
> two AAAA records and my IPv6 connectivity is subtly broken?

This is really the wrong mailing list for this kind of discussion; I had
started IPv6 implementation a while ago on time-dev.

You are right; DNS is the first issue. It starts with the problem that
the DNS server used for pool.ntp.org doesn't even support AAAA records.
It's pgeodns, and I created a patch to support IPv6 a while ago.

The next issue is the UI; I have patches for this as well.

Then the next issue is the monitoring of servers; I am working on patches.

Only *then* the next issue is the exact deployment procedure. I propose
that there is a single zone ipv6.pool.ntp.org first - both because there
will be initially only a small number of servers, and because of
possible client problems, too. So using IPv6 would be an opt-in
decision. Before even creating the zone, number of good-quality
servers need to be in the pool.

Once there is a significant number of servers registered for IPv6,
adding them to the existing zones can be considered. At that point,
likelyhood of bad IPv6 connections on clients can be considered
(and potentially responded-to with an opt-out zone ipv4.pool.ntp.org).

Regards,
Martin
_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to