BIND 9.2.1 MINTTL Problem (and Solution)
Danny Mayer
mayer at gis.net
Sat Jul 6 14:19:05 UTC 2002
At 05:14 PM 7/3/02, Jim Reid wrote:
> >>>>> "Barry" == Barry Finkel <b19141 at achilles.ctd.anl.gov> writes:
>
> Barry> I had not read the migration notes, but I knew about the
> Barry> $TTL directive, which I have in all of my master zones. I
> Barry> read the migration notes, and these notes would not have
> Barry> helped me with this problem.
>
>They should have made it straightforward to figure out the problem. The
>BIND9 server loaded old zone files that did not have a $TTL directive
>or an explicit TTL on the SOA record. That's what caused the warning
>message in the logs. Those slave zone files didn't have these things
>because they were old, perhaps dating from the time before BIND fully
>supported RFC2308 semantics. So the question then becomes how to get
>new zone files on the slave server(s) that include the missing TTL: ie
>a forced transfer for each of those zones. If you'd moved those old
>files out of the way before starting named, the server would have
>performed zone transfers and then written out fresh copies of the zone
>files. These would include the missing TTL, which would shut up the
>server the next time it was restarted and loaded those zone files.
The proper way would have been to just increment the serial number and
the slaves would have pulled a new copy of the zone with the right
semantics.
Danny
More information about the bind-users
mailing list