[ntp:questions] Bad NTP servers jeopardizing the pool.ntp.org initiative
mayer at ntp.isc.org
Mon Mar 26 19:20:58 UTC 2007
Wolfgang S. Rupprecht wrote:
> hal-usenet at ip-64-139-1-69.sjc.megapath.net (Hal Murray) writes:
>> I'm sure what the right heuristics are. Maybe something
>> like once-a-month and if DNS fails, keep using the old address
>> as long as it works.
> It has always bothered me that the TTL in the DNS record is stripped
> off by the resolver code. There is already an indication in the DNS
> record how long the issuer of the record wants it to be used. Careful
> DNS admins will even go so far as creeping the TTL down as a cut-over
> to a new IP address happens. It would be ideal for a long-running
> program to just time out the IP address using this TTL and request a
> new one whenever the old one expires.
The data structure addrinfo used for getaddrinfo() would need to be
modified to do this in a useful way. But you should look at subsequent
addresses and not just the first one as well.
> Seeing how ntpd is now hosted at ISC, anyone here have the ear of the
> bind developers?
I was one of them. But please remember that while NTP Public Service
Project is *hosted* at ISC does not mean that it is part of ISC. It isn't.
More information about the questions