[Pdns-users] Multiple Masters causing strange issue

[Stephen Harker]

Hello again,

I've successfully upgraded our DNS to the latest version of PowerDNS - 
2.9.22 - and got the OpenDBX/MySQL and auto-serial stuff working nicely. 
However I now have a new problem.

We have several hundred slave domains configured for various different 
customers and each customer has their own Master server (Bind usually) 
on a separate IP address and they manage their zone files themselves.

What's happening now is that PowerDNS seems to be munging all of those 
different slaves in to a pool and trying all of them in turn, over and 
over in a loop when it encounters a problem with a domain. So trying to 
get the SOA for a domain from a server that's in most cases not 
authoritative for that domain and causing all sorts of grief and a lot 
of DNS traffic. It does this many times a second for each domain and 
eventually gives up and moves on to the next. I would expect a 
problem/timeout loading an SOA to get the serial for a slaved domain to 
cause nothing more than for PowerDNS to move on to the next domain, 
instead it's getting in to a frantic state!

Does anyone know what might be causing this?

Here is the domains entry for a particular domain:

mysql> select * from domains where name='somedomain.com'\G
*************************** 1. row ***************************
              id: 49934
            name: somedomain.com
          master: 12.34.131.163
      last_check: 1248691024
            type: SLAVE
notified_serial: NULL
         account: 1234567890-A-BCD
          status: A
     auto_serial: 0
1 row in set (0.00 sec)

mysql>

And a sanitised example of the log entries (they are copious)

Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Skipping query to '12.34.128.229' because of previous 
timeouts in this cycle
Mar 17 17:26:22 dnsserver pdns[15018]: Trying next master 
'12.34.131.153' for 'somedomain.com '
Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Skipping query to '12.34.131.153' because of previous 
timeouts in this cycle
Mar 17 17:26:22 dnsserver pdns[15018]: Trying next master '12.34.128.11' 
for 'somedomain.com '
Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Skipping query to '12.34.128.11' because of previous 
timeouts in this cycle
Mar 17 17:26:22 dnsserver pdns[15018]: Trying next master 
'12.34.128.208' for 'somedomain.com '
Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Query to '12.34.128.208' for SOA of 'somedomain.com ' 
produced a NS record
Mar 17 17:26:22 dnsserver pdns[15018]: Trying next master 
'12.34.128.106' for 'somedomain.com '
Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Query to '12.34.128.106' for SOA of 'somedomain.com ' 
produced a NS record
Mar 17 17:26:22 dnsserver pdns[15018]: Trying next master 
'12.34.131.106' for 'somedomain.com '
Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Skipping query to '12.34.131.106' because of previous 
timeouts in this cycle
Mar 17 17:26:22 dnsserver pdns[15018]: Trying next master 
'12.34.128.148' for 'somedomain.com '
Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Skipping query to '12.34.128.148' because of previous 
timeouts in this cycle
Mar 17 17:26:22 dnsserver pdns[15018]: Trying next master 
'12.34.131.153' for 'somedomain.com '
Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Skipping query to '12.34.131.153' because of previous 
timeouts in this cycle
Mar 17 17:26:22 dnsserver pdns[15018]: Trying next master 
'12.34.128.105' for 'somedomain.com '
Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Skipping query to '12.34.128.105' because of previous 
timeouts in this cycle
Mar 17 17:26:22 dnsserver pdns[15018]: Trying next master 
'12.34.128.106' for 'somedomain.com '
Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Query to '12.34.128.106' for SOA of 'somedomain.com ' 
produced a NS record
Mar 17 17:26:22 dnsserver pdns[15018]: Trying next master 
'12.34.128.106' for 'somedomain.com '
Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Query to '12.34.128.106' for SOA of 'somedomain.com ' 
produced a NS record
Mar 17 17:26:22 dnsserver pdns[15018]: Trying next master 
'12.34.131.153' for 'somedomain.com '
Mar 17 17:26:22 dnsserver pdns[15018]: Error trying to retrieve/refresh 
'somedomain.com ': Skipping query to '12.34.131.153' because of previous 
timeouts in this cycle

Kind regards,

Stephen Harker

-- 
All postal correspondence to: Positive House, 24 Broadway, London. W13 0SU

The Positive Internet Company Limited is registered in England and Wales.
Registered company number: 3673639. VAT no: 726 7072 28.
Registered office: Northside House, Mount Pleasant, Barnet, Herts, EN4 9EE.