[Pdns-users] Recursor crashes

[Tobias Orlamuende]

Hi,

since today from around 1:30 pm we are having recursor crashes on both of our nameservers.
They are running pdns-recursor 2.9.17-13sarge3 and pdns-server 2.9.17-13sarge3 (as you can see, both on Debian).

Config on those machines is, that pdns-server listens on the inet-inbterface on port 53 tcp and udp.
Recursor listens on 127.0.0.1 port 53 tcp and udp.

Both machines were running for about one year without any problems or crashes. This afternoon I did an update of the pdns-packages, but the error does not have anything in common with that update since the problems occured with the older version as well.

Logs only show:

Dec 20 16:25:34 ns2 pdns[2586]: Recursive query for remote XXX with internal id 37 was not answered by backend within timeout, reusing id

Where XXX is the IP of one of our servers.

and

Dec 20 16:29:40 ns2 pdns[2586]: Error receiving packet from recursor backend: Connection refused

DB-backend is gmysql with connects to localhost and MySQL-Replication.

Currently I do some debugs on one machine and since I startet this, it does not crash anymore, but the other does.
The debugs I do are to strace the pdns-recursor-process and tcpduming incomming connect to port53.

Since these machines were running very well, absolutely nothing was changed at the configs or applications.
Can anybody help with that strange problem?
TIA!

Cheers

Tobias