Question on documentation (TTL or not TTL, that is the question....)

Cricket Liu cricket at VeriSign.com
Sun Mar 4 21:36:12 UTC 2001


> Question on documentation....
>
> I've been reading DNS&BIND 3rd edition (I understand the 4th is not yet
off
> the presses). There seems to be a bit of a clash in the SOA record
> definition.
> Did it change from BIND8 to BIND9 (can't imagine)?

No, it changed between BIND 8.1.2 and 8.2, when RFC 2308
was published.

> On p.90 the Book says that the last value in the SOA record is the
'minimum
> TTL'
> which is described as the minimum time a record lives in a caching
server's
> cache (which is also what http://www.ietf.org/rfc/rfc1035.txt says)

That edition of the book covers BIND 8.1.2.

> The BIND9 documentation says that the last field in the SOA is the
negative
> caching TTL, which controls how long other servers will cache NXDOMAIN
> responses from the server. (This can be found in paragraph 6.3.3 of the
> admin manual)

That's currently correct, and what you'll find in the fourth edition of
the book.

> Slave BIND 9 name servers create database files which are commented like
> this, thus giving credence to RFC1035:
>
> taxi.lu                 IN SOA  dns1.synapse.lu. dnsadmin.synapse.lu. (
>                                 2001020501 ; serial
>                                 28800      ; refresh (8 hours)
>                                 7200       ; retry (2 hours)
>                                 604800     ; expire (1 week)
>                                 3600       ; minimum (1 hour)
>                                 )

That must be an oversight.

cricket



More information about the bind-users mailing list