[dns-operations] NSATC DNS oddities last week (affecting hotmail, msn etc)

[Florian Weimer]

* bert hubert:

> 2) It has been suggested that PowerDNS invalidate an NS record from its
> cache if it is not responsive, and refetch from higher up. I recall BIND
> implemented something like lowering the TTL quicker than realtime based on
> query load, which would implement something like this.
>
> Is this considered a good idea?

I've seen query storms to the .COM servers which could be attributed
to this behavior in a questionable implementation.  Generally capping
TTLs seems safer.