NS record TTL versus nameserver's A record TTL
jw354 at cornell.edu
Tue Oct 8 15:49:24 UTC 2013
We received a report that a domain we serve
was not resolving at a remote site. The site
also reported their own analysis that the issue
appeared to be that the domain's NS record had
a longer TTL than its target nameserver's
A record and their caching server didn't
seem able to handle this. FYI, the nameserver
was not within the domain with the issue.
They took responsibility for their
nameserver's deficiency, but it
makes me wonder:
-Is this addressed by a standard? E.g.,
the nameserver's A record have the same
TTL as NS records pointing at it.
-Is this addressed by a "best practice"?
-If neither of the above, is there
a "hidden practice that knowing folk
often follow to dodge remote
FYI, I only received the report fourth hand
and can't tell you the nameserver software
that had this issue.
Cornell University IT
P.S. This made me wonder what record bind
puts in the additional section if it has
both a glue A record for a nameserver
in the zone's file and an authoritative A
record for the nameserver in the nameserver's
own zone file. I find by TTL finagling that
it serves the authoritative A record.
More information about the bind-users