[dns-operations] Also any Microsoft CDN people here?

[Gavin McCullagh]

On Thu, Nov 27, 2025, 2:12 PM Peter Thomassen <peter at desec.io> wrote:

> Of course. I think the issue here is that the reproducible switching
> between existence and non-existence, while there is a also caching, is very
> unlikely to be the result the zone maintainer had in mind.
>

This makes sense alright.  It does seem like it may not result in the
outcome the zone owner wants.

It's not "invalid", protocol-wise, but it's probably "wrong" anyway (in the
> bug sense).
>

That's the distinction I was looking to draw alright.

If this truly causes problems for a resolver, that seems a different
problem as this exact behavior can happen (much less frequently) with any
nameserver, due to changes made within a zone.   In the face of conflicting
cacheable responses, I assume (I have not implemented a resolver myself) a
resolver has to choose the more recent response on the assumption the zone
may have changed.

Gavin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.dns-oarc.net/pipermail/dns-operations/attachments/20251127/5ff4fb21/attachment-0001.html>