[Pdns-users] supermaster sends notify to itself on zone update?

[other at ahhyes.net]

Hi Guys,

I am seeing some odd entries on my master server logs (same behavior 
witnessed in 3.1 and 3.2-rc4).

Log Snippet:

Jan 22 13:01:32 ns pdns[53002]: 1 domain for which we are master needs 
notifications
Jan 22 13:01:32 ns pdns[53002]: Queued notification of domain 
'example.net' to 2001:ip6:ip6:ip6::ip6
Jan 22 13:01:32 ns pdns[53002]: Queued notification of domain 
'example.net' to 212.ip.ip.ip
Jan 22 13:01:32 ns pdns[53002]: Queued notification of domain 
'example.net' to 2607:ip6:ip6:ip6::ip6:ip6
Jan 22 13:01:32 ns pdns[53002]: Queued notification of domain 
'example.net' to 65.ip.ip.ip
Jan 22 13:01:32 ns pdns[53002]: Received NOTIFY for example.net from 
2001:ip6:ip6:ip6::ip6 but slave support is disabled in the configuration
Jan 22 13:01:33 ns pdns[53002]: AXFR of domain 'example.net' initiated 
by 2607:ip6:ip6:ip6::ip6:ip6
Jan 22 13:01:33 ns pdns[53002]: AXFR of domain 'example.net' allowed: 
client IP 2607:ip6:ip6:ip6::ip6:ip6 is in allow-axfr-ips
Jan 22 13:01:33 ns pdns[53002]: AXFR of domain 'example.net' to 
2607:ip6:ip6:ip6::ip6:ip6 finished
Jan 22 13:01:33 ns pdns[53002]: Received unsuccessful notification 
report for 'example.net' from [2001:ip6:ip6:ip6::ip6]:53, rcode: 4
Jan 22 13:01:33 ns pdns[53002]: Received spurious notify answer for 
'example.net' from [2001:ip6:ip6:ip6::ip6]:53
Jan 22 13:01:33 ns pdns[53002]: Received spurious notify answer for 
'example.net' from [2607:ip6:ip6:ip6::ip6:ip6]:53
Jan 22 13:01:33 ns pdns[53002]: Received unsuccessful notification 
report for 'example.net' from 212.ip.ip.ip:53, rcode: 5
Jan 22 13:01:33 ns pdns[53002]: Removed from notification list: 
'example.net' to 212.ip.ip.ip:53
Jan 22 13:01:35 ns pdns[53002]: Received NOTIFY for example.net from 
2001:ip6:ip6:ip6::ip6 but slave support is disabled in the configuration
Jan 22 13:01:35 ns pdns[53002]: No master domains need notifications

Firstly, everything appears to be working ok, updates to the slave work 
fine. I am just puzzled by this bizarre way powerdns does things, the 
first 2 lines of "Queued notification" events are queuing notifications 
to the master itself, why? This is the reason for the spewing of some of 
the "Received unsuccessful notification", "Received spurious notify 
answer" and specifically "slave support is disabled in the 
configuration" - slave operation is not required on the supermaster.

Does anyone have any insight into why this occurs/why powerdns does 
things this way? The domain itself involved in the log has 2 ns records 
in its zone, which resolve to the ip of the master and the ip of the 
slave respectively (both via ipv4 and ipv6).

Cheers,
Alex.