[Pdns-users] version/AXFR question

Dave Taylor davetaylor at frontiernet.net
Fri Sep 23 20:15:45 UTC 2005


We're seeing a rather minor issue but haven't been able to figure out why.
We have 1 test server running version 2.9.17-13sarge1 and 1 production
server running 2.9.17-13 both on Debian.  In the pdns.conf file for both
machines, disable-axfr=yes is set (uncommented in the config file).  Both
servers also have slave=yes set.  

 

In our setup, any domain in which we are the primary nameserver gets entered
in the DB as NATIVE and mysql replication takes care of updating the 2nd
production nameserver.  Any domain for which we are only a secondary
nameserver gets entered as SLAVE in the DB and the domain's master will
notify us of changes, initiating the AXFR.  

 

What we noticed is that if we change a domain from NATIVE to SLAVE and give
it a new master IP, within 1 minute the test server notices that 1 slave
zone needs updating and it attempts to do the AXFR without any intervention
from us.  This is how I would assume it would work.  But on the production
server, when a domain is changed to a SLAVE, nothing happens unless we run
pdns_control retrieve domain_name from the command line.

 

It would also be good to point out that we are making these changes via our
own home grown web based front end.  If a domain gets changed from NATIVE to
SLAVE, the serial is reset to 0 in the SOA record.

 

What exactly does the disable-axfr setting do?? And are we missing something
simple here?

 

Thanks,

Dave 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.powerdns.com/pipermail/pdns-users/attachments/20050923/b9e64607/attachment.html>


More information about the Pdns-users mailing list