BIND 9.9.1-P1 reload bug

[Cathy Almond]

> This just happened on our nameserver:
> 
> 11-Jul-2012 13:54:01.711 general: info: received control channel command
> 'reload'
> 11-Jul-2012 13:54:01.712 general: info: loading configuration from
> '/etc/named.conf'
> 11-Jul-2012 13:54:01.891 general: critical: server.c:4436: fatal error:
> 11-Jul-2012 13:54:01.891 general: critical: RUNTIME_CHECK(result == 0)
> failed
> 11-Jul-2012 13:54:01.891 general: critical: exiting (due to fatal error
> in library)

I think this is a fairly-recently identified bug (#29872).  It's a race
condition between management of the Address Database part of cache (ADB)
and the rndc reload.  There isn't a patch available for it, but it will
be fixed in a future release.

It's also more likely to occur after a rndc reload on a server that's
been fairly recently restarted and hasn't yet reached a stable-state
cache population - although it still can happen some significant time later.