Questions on dynamic DNS nsupdate command

[Pozar, Frank]

Have a couple questions regarding the transition from statically maintained
dns zones (using hosts_to_named script/h2n scripts )  to  dynamically
controlled zones using nsupdate...

1.When I issued the nsupdate command the first time on a zone, the format of
the zone file on disk (ex db.new) was changed. It put a ;BIND DUMP V8 at the
top of the file and appended a ;Cl=5 at the end of each line. Also, the
numbers for the serial no. , expire , refresh and ttl times were all placed
on one line.
Is this file corrupted or is it normal for a dynamically controlled zone
file to have this format?

2. How does  a dynamic zone get loaded into memory upon startup of the named
daemon on the nameserver.... What files are read.. Does it read the zone's
actual zone file on disk and then the .log file, if it exists, and puts it
into memory?  If I want to backup the zone files I should backup these fies
for disaster recovery?

3. I want to restate the obvious for confirmation... once a zone is being
dynamically updated via the nsupdate command or a DHCP routine is hitting
it, it should not be modified by hand (vi"ing" the zone incrementing the
serial number.. or running h2n).

4. We are trying to implement Windows 2000 seats on a subnet owned(authority
given) by a UNIX nameserver. So the situation we are facing is the the
reverse zones have both unix and windows clients in it while the forward
zones are owned by the respective name servers (the windows 2000 domain
controller running DNS controls the forward zone for the windows 200 seats
while the unix server owns the forward zone for unix seats).

So for the shared reverse zone, we would have dynamic updates occuring via
nsupdate command from the unix side and via DHCP controller from the windows
2000 side.. Has anyone implemented such a setup?????  Any suggestions...

On the unix side, we used the host_to_named script to pull changes in the
NIS host map into the dns zone files..  Is there any script using the
nsupdate command to perform dns updates which would replace the
host_to_named  process?????

Would there be an issue if a nsupdate from a unix system was running on a
zone at the same time a windows DHCP controller was submitting dynamic
updates via there ns_update routine?????

5. Is the current suggested roadmap to have the windows 2000 clients on a
separate subnet from UNIX clients??  Therefore, the window systems are
taking control of both the forward and reverse domains.. and unix takes care
of is own...

6. For bind version 8.2.2 (BIND V8), does the serial number change for a
zone dynamically updated occur about 5 minutes after the dynamic change to
the zone in memory??? Therefore the secondary or slave nameservers would not
see this update for 5 minutes??? Is this true???   


Thanks in advance for all your help..
Frank