[Pdns-users] PowerDNS does not start if one of the local-address is down.

Augie Schwer augie.schwer at gmail.com
Sat Nov 4 01:21:28 UTC 2006

If one of the listed address in "local-addresses" is down, then
PowerDNS spins and does not bring up any of the good addresses until
it can bring up all of the addresses.

For example; local-addresses=, ; and say the interface
that is bound to is down, then PowerDNS will cycle until it
is back up; the log looks something like this:

Nov  3 16:47:20 a pdns[6624]: Guardian is launching an instance
Nov  3 16:47:20 a pdns[6624]: This is a guarded instance of pdns
Nov  3 16:47:20 a pdns[6624]: UDP server bound to
Nov  3 16:47:20 a pdns[6624]: binding to UDP socket: Cannot assign
requested address
Nov  3 16:47:20 a pdns[6624]: Fatal error: Unable to bind to UDP socket
Nov  3 16:47:21 a pdns[6616]: Our pdns instance exited with code 1
Nov  3 16:47:21 a pdns[6616]: Respawning

This type of behavior is problematic for our current deployment; where
the order of operation with BIND has been to first do a health check
locally for zones, and if the health is deemed good, then bring up
several interfaces and announce those so traffic can then be routed to
the box.

We can probably do some workarounds, but it does seem like a bug for
PowerDNS to fail if it cannot bind to all of the addresses listed in
local-addresses. Ideally it seems that PowerDNS should bind to the
addresses it can, and then re-scan at some interval for availability
of the other addresses and then bind to those when they are available.

Augie Schwer    -    Augie at Schwer.us    -    http://schwer.us
Key fingerprint = 9815 AE19 AFD1 1FE7 5DEE 2AC3 CB99 2784 27B0 C072

More information about the Pdns-users mailing list