Bind 9 slaves - new zones are never picked up

[mhageman]

Hi,

I am migrating from bind 8.2.4 to bind 9.  Our house master and all but
two slave are running bind 8.2.4.  I replaced one of the main
production slaves with an AIX 5.2 box running 9.2 last week.

The new slave, and a backup AIX 5.1 box running 9.2.1 are performing
well except in one area:  we add and delete zones regularly because we
are a "project" support site and we have customers with different name
and IP zones coming-and-going regularly.  The bind 9 production (and
backup) box NEVER transfer/service the new zones.    rndc requests
don't help.  refresh -s named doesn't help.  I've had to resort to
kill-9 and restarting (TERM takes too long as the servers are always
very busy).  Needless to say, this is completely unacceptable.

Our procedure for updating DNS is a much-modified "h2n" within a shell
that runs on the master and "rdists" the changed named.conf files to
our slave servers.  Worked great on V4.  Works great on V8 - the slaves
AXFR the new zones pretty quickly.

Doesn't work at ALL on bind 9.  So I am missing something, and it must
be something in the configuration.

Can anybody help me? 

Thanks,

Mary