NOERROR when NXDOMAIN expected

Jim Reid jim at rfc1035.com
Fri Mar 2 09:32:16 UTC 2001


>>>>> "Rainer" == Rainer Ginsberg <rainer.ginsberg at de.bosch.com> writes:

    >>  IIRC the RFC2308 behaviour of BIND can be tweaked with the
    >> rfc2308-type1 option.

    Rainer> Strange enough, but my BIND 8.2.3-REL doesn't change it's
    Rainer> behaviour to type 1 (i.e. sending NS records along with
    Rainer> the SOA record in the authority section) when setting
    Rainer> rfc2308-type1 to yes. Also, "has-old-clients no;" has no
    Rainer> effect.

Strange. Can you provide more details, eg sample dig output?

    >> I suppose you could also fake a reply with a TTL longer than
    >> the M$ server's uptime for these non-existent RRsets. That
    >> would shut them up too. :-)

    Rainer> I don't think so. Microsoft is ignoring TTLs for negative
    Rainer> caching.  Instead, they use two registry keys, as far as I
    Rainer> know (one for SOA records and one for other record types).

Who said those faked answers had to be negative answers? :-)


More information about the bind-users mailing list