Is it possible to tell when a record will expire?

[Barry Margolin]

In article <3.0.3.32.20000614233205.01d5e164 at pop3.hank.org>,
Bill Moseley  <moseley at hank.org> wrote:
>I added a machine to my zone file about 24 hours ago.  Of the accounts I
>have access to on different machines almost all resolved this new machine's
>IP number.
>
>One check didn't have that new record, though.  So I have two questions:
>
>1) I assume the RRs are cached and not the entire zone.  So if I queried
>for a new addition to my zone file (that has also been transferred to all
>slaves) the local DNS will query an authoritative server for the info.
>Correct?
>
>2) I also assume if I changed the IP of an existing name that I would have
>to wait for the local cached RR to expire before it would query the
>authoritative server for the new info.  So, is there any way to look at the
>local server with nslookup or dig to find out how much time is left until
>the record expires?

Dig, or NSLOOKUP with "set debug", shows the TTL of the records.  That's
how long it has until it expires (TTL = "time to live").

>Oh, BTW, I just looked some more and may have found the reason the new name
>didn't resolve (although I'm still curious about the above questions).
>
>It's probably a granitecanyon problem.  Yesterday ns1.granitecanyon.com
>wasn't answering queries for my domain -- it was referring to the root
>servers instead.  But now it is answering queries for my domain, yet,
>according to the SOA record it has, is using an old copy of my zone file.
>Humm.  Seems odd that it would drop my record and then later (after I
>changed my zone on my primary and sent out a NOTIFY) that it would start
>answering for my domain, yet with an old zone record.
>ns2.granitecanyon.com has the correct info.  Odd.

I suspect that ns1 somehow lost its named.conf file yesterday (it was
also missing a zone that someone else was asking about), and then it got
put back.  Perhaps as a result of that it has gotten out of date on lots of
zones, and hadn't caught up yet in transfering them.

BIND also has a bug where a zone gets into a state where it thinks it's in
the middle of transfering it, but it's not.  But because this flag is set,
when the refresh time comes around again (or a NOTIFY is received from the
master) it doesn't try transfering again.  Sometimes one of our servers
will end up with hundreds of domains in this state, and the only fix we've
found is "ndc restart".

-- 
Barry Margolin, barmar at genuity.net
Genuity, Burlington, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.