Hi Chris,

On 12/15/21 10:28 PM, Chris wrote:
> On 2021-12-15 13:01, Anand Buddhdev wrote:
>> On 15/12/2021 20:18, Chris wrote:
>>
>> Hi Chris,
>>
>> [snip config details]
>>
>>> How would I best make this change? Is it enough to simply change algorithm:
>>> and knot will just do the right thing?
>>
>> Yes, please! Just change the algorithm and let Knot do its thing. It will do 
>> the
>> right thing. Please do *not* fiddle with things manually. DNSSEC is complex, 
>> and
>> algorithm roll-overs require care. The developers of Knot have put in a lot 
>> of
>> care into handling algorithm roll-overs. Trust their expertise.
> Thanks for the reply, Anand! :-)
> I'm well aware of all the complexities, and am well confident in knots 
> abilities
> to DTRT. But "stuff" happens. fe; after creating the additional policy
> some of the zones are _also_ adopting that new policy as _well_ as the 
> original
> policy. IOW there are some zones with both RSASHA1 _and_ RSASHA256 hashes in 
> them.

One zone cannot use more DNSSEC policies! I think you are confused by ongoing 
algorithm
rollover when there are both algorithms present in the zone (see 
https://datatracker.ietf.org/doc/html/rfc6781#section-4.1.4).

> 
> config (diffs):
> policy:
>   - id: rsa1
>     algorithm: RSASHA1
>     zsk-size: 1024
> 
> policy:
>   - id: rsa2
>     algorithm: RSASHA256
>     zsk-size: 2048
> 
> ALL zones but the test zone mentioned earlier:
> 
>   - domain: domain.name
>     ...
>     dnssec-signing: on
>     dnssec-policy: rsa1
> 
> So why do (some) zones arbitrarily pick up the added policy when it
> it is not the policy declared within the domain block?

Isn't it possible that the policy is declared in a zone template?

Daniel


> IOW dnssec-policy: rsa1 is the only dnssec-policy listed within all the
> domain blocks, and it's listed within all of the domain blocks, save
> the earlier test domain. So "stuff" happened. :-/
> 
> Thanks again, for taking the time to respond, Anand.
> 
> -- Chris
>>
>> Regards,
>> Anand
-- 
https://lists.nic.cz/mailman/listinfo/knot-dns-users

Reply via email to