DNSSEC and child zones on same authoritative NS. Expert help needed.
Gary Wallis
wgg1970 at gmail.com
Tue Mar 16 16:54:38 UTC 2010
Sam Wilson wrote:
> In article <mailman.814.1268703621.21153.bind-users at lists.isc.org>,
> Gary Wallis <wgg1970 at gmail.com> wrote:
>
>> Let's say I have this setup :
>>
>> BIND 9.4 named.conf includes a master.zones file with the following:
>>
>> ...
>> zone "ns1.yourdomain.com" {
>> type master;
>> file "master/external/n/ns1.yourdomain.com.signed";
>> };
>>
>> zone "ns2.yourdomain.com" {
>> type master;
>> file "master/external/n/ns2.yourdomain.com.signed";
>> };
>>
>> zone "yourdomain.com" {
>> type master;
>> file "master/external/y/yourdomain.com.signed";
>> };
>> ...
>>
>> More background for question below:
>>
>> The yourdomain.com is I gather the zone APEX for all featured zones
>> above. (Is this the correct use of the term APEX?)
>
> "Parent", as Mark has already pointed out.
Got that :)
I would be nice to know what a zone apex is since what I have found on
the web so far is pretty self-referential.
>
>> I am learning via trial and error about transitioning from DNS to DNSSEC
>> and we have these child zones (is ns1.yourdomain.com really a child
>> zone, as regards the setup above?) that currently have precedence over
>> the parent zone yourdomain.com for conflicting A records. For example:
>>
>> If
>>
>> ns1 A 123.123.123.123
>>
>> is placed in yourdomain.com zone.
>
> Some nitpicking - I'm not a DNSSEC expert and I'm not commenting on that
> part of your question. Including this record would normally be an
> error. ns1.yourdomain.com is delegated into its own zone and the A
> record should be in that zone, not in the parent zone.[1]
>
>> And a similar RR is placed in ns1.yourdomain.com zone, like:
>>
>> ns1 IN A 10.0.0.1
>
> If you place ns1 in the zone ns1.yourdomain.com then the name will be
> ns1.ns1.yourdomain.com. If you force the name to be ns1.yourdomain.com
> [2] then that A record should override the one in the parent zone (see
> [1] again).
Good job! You spotted my error/typo the test setup actually is:
...
@ IN A 10.0.0.1
...
for ns1.yourdomain.com zone file
>
>> And named reloaded.
>>
>> dig @localhost ns1.yourdomain.com A +short
>>
>> will return 10.0.0.1, the parent A RR is ignored.
>
> Correct - see above
>
> Can't answer your DNSSEC queries, but I'm not sure if they're relevant
> if you correct the above.
Regarding my main question:
How to delegate signing authority from parent yourdomain.com to child
ns1.yourdomain.com.
In this case in the same BIND configuration (same named daemon) as shown
above in the named.conf frag.
I also need to know if the initial zone signing has to be changed, i.e.
sign the parent and child zones differently etc.
It seems that I need to provide the child DS record from
dsset-ns1.yourdomain.com that was generated by dnssec-signzone when I
signed the child.
dsset-ns1.yourdomain.com contents:
ns1.yourdomain.com. IN DS 181 5 1
FD110AAAFAC8101DD8EC946FD5B62FDC9B012EA1
That would go in the yourdomain.com parent db file. But I imagine some
other steps are needed and I'm not sure what they are.
I am reading about DS records now. But if anyone has a short how-to (or
listing of bash commands) for this simple case that would be great.
I still have to setup a DNSSEC resolver to be able to test my "test"
auth DNS server. And will provide results and a how-to if I manage to
get this "internal" DNSSEC parent-child delegation working on my own.
>
> Sam
>
>
> [1] UNLESS ns1.yourdomain.com is also the name of one of the nameservers
> for a child zone in which case that record would be a glue record which
> would be valid in the parent zone. It would normally be superseded by
> the corresponding A record in the child zone which is regarded as a more
> trustworthy source of data. There are various ways by which a server for
> the parent zone can learn the correct data from the child zone.
This is what I know from experience but have never seen written
anywhere. Thx!
>
> [2] You can do that by using the @ sign in the LHS of the A RR, or by
> using a fully qualified name (inflexible), or by using the $ORIGIN
> directive, or by leaving the name blank at the head of the zone
> (slightly risky). Of these @ is the one mostly recommended.
Good. We started using @ several year ago but did not know about all
these details and differences you mention.
> _______________________________________________
> bind-users mailing list
> bind-users at lists.isc.org
> https://lists.isc.org/mailman/listinfo/bind-users
>
Thanks all!
More information about the bind-users
mailing list