Minimum TTL?
Reindl Harald
h.reindl at thelounge.net
Fri Feb 9 17:08:01 UTC 2018
Am 09.02.2018 um 17:45 schrieb Barry Margolin:
> In article <mailman.452.1518194404.749.bind-users at lists.isc.org>,
> Reindl Harald <h.reindl at thelounge.net> wrote:
>
>>> As long as you understand the implications of what you're doing?
>>>
>>> The zone owner may be using short TTLs to implement load balancing
>>> and/or quick failover. If you extend the TTLs, your users may experience
>>> poor performance when they try to go to these sites using out-of-date
>>> cache entries
>>
>> but that's my problem then and not yours - it's that simple
>
> Sure, but the Internet was designed on a philosophy of cooperation. An
> ISP could also drop every other packet, and say "that's my problem, not
> yours", but we wouldn't consider that to be a reasonable way to run a
> network.
>
> IMHO you should at least be transparent about it, so your users know
> what they're in for
where i would place that option "my users" are my servers (inbound MX,
RBL's hence unbound there, but you would know that if you would have
followed the thread)
another usecase are 5 seconds or so to mask problems of the zone-owner
where all his slaves are victims of Cisco hardware and mangle CNAMEs in
zone-transfers with a "$TLL 0" in front of them while the whole domain
was intened to have a global 86400 seconds TTL
one needs me to show a single example where human users would have a
non-theoretical differnece between 2 and 5 seconds..
but you would also know that if you have followed the thread
More information about the bind-users
mailing list