unbound flooding syslog with 'sendto failed: Cannot assign requested address'

Taylor R Campbell campbell+unbound at mumble.net
Fri Oct 16 13:22:26 UTC 2015


   Date: Fri, 16 Oct 2015 13:28:12 +0200
   From: "W.C.A. Wijngaards via Unbound-users" <unbound-users at unbound.net>

   On 10/16/2015 12:47 PM, Hendrik Jaeger via Unbound-users wrote:
   > I'm running unbound 1.4.22 from debian unstable on my laptop. Since
   > yesterday unbound started behaving strangely, flooding the syslog 
   > with failure messages. Example: Oct 16 10:23:20 frustcomp unbound:
   > [26958:0] notice: sendto failed: Cannot assign requested address 
   > Oct 16 10:23:20 frustcomp unbound: [26958:0] notice: remote address
   > is 2001:500:3::42 port 53

   The newer versions of unbound have many fixes.  Several of them have
   related to squelching log messages.

On many versions of Unbound, on NetBSD 5 and 6, when the network goes
down, whether or not IPv6 is involved, I have seen it spew essentially
this message to the log, e.g.:

[1444927008] unbound[27554:0] notice: remote address is 192.228.79.201 port 53
[1444927008] unbound[27554:0] notice: sendto failed: Can't assign requested address

However, even if the log is sent to /dev/null, Unbound is spinning
trying and failing to do sendto as fast as it can and draining my
laptop battery.  So I'm worried that `squelching log messages' may
paper over a legitimate error here, and I wonder whether any bugs
involving loops with sendto -- whether or not logging -- have been
fixed.



More information about the Unbound-users mailing list