[bind10-dev] unknown SERVFAIL

[JINMEI Tatuya / 神明達哉]

At Mon, 22 Aug 2011 09:59:32 -0500 (CDT),
"Jeremy C. Reed" <jreed at isc.org> wrote:
> 
> On a new startup, my first query was for whitehouse.gov and received 
> SERVFAIL. I got a bunch of:
> 
> (09:11:31) jreed: 2011-08-05 08:51:38.945 ERROR [b10-resolver.asiolink] 
> ASIODNS_SEND_DATA error 65  sending data using UDP to 2001:503:a83e::2:31(53)
> 
> 2011-08-05 08:51:38.965 ERROR [b10-resolver.nsas] NSAS_INVALID_RESPONSE 
> queried for d.gtld-servers.net. but got invalid response
> 
> Then did a dig for isc.org. That worked. Then immediate dig for 
> whitehouse.gov again and that worked.  From this logging I can't really 
> know what caused this.  65 is EHOSTUNREACH (No route to host)
> 
> So that system has IPv6 but not IPv6 network. Maybe it didn't or 
> couldn't fallback to IPv4?

I've not checked the code see whether it falls back to other addresses
(I believe it does though), but in any case ERROR is IMO too strong for
this type of event.  "No route to host" can happen under normal
situation and it may be specific to a particular set of addresses we
cannot control.  So the ERROR messages could be unnecessarily noisy.

We should also avoid hardcoding the magic number in the log; it should
be converted to a human readable string by strerror() or something (I
guess at the level of asiolink we can use a utility of ASIO).

---
JINMEI, Tatuya