Hi Petr,

the ZSK rollover with the two keys having the identical key tag went fine.
The only thing that was not so nice was that I could not follow the roll
over progress in the log file. In the log I am not able to see which key
is which.

Not sure if it's worth the effort to extend the source code to avoid
unique keys. I think this a very rare situation.


Best,
Thomas

On 07.07.20 09:23, Petr Špaček wrote:
> Hi everyone,
> 
> let me add references:
> https://tools.ietf.org/html/rfc4034#appendix-B explains this:
>>    However, it is essential to note that the key tag is not a unique
>>    identifier.  It is theoretically possible for two distinct DNSKEY RRs
>>    to have the same owner name, the same algorithm, and the same key
>>    tag.  The key tag is used to limit the possible candidate keys, but
>>    it does not uniquely identify a DNSKEY record.  Implementations MUST
>>    NOT assume that the key tag uniquely identifies a DNSKEY RR.
> 
> Also key tag "collision" is even more likely than it seems at first glance, 
> see 
> https://lists.dns-oarc.net/pipermail/dns-operations/2015-December/013949.html
> 
> One question is if generating keys in a way which quarantees unique key tags 
> would improve user experience sufficiently to warrant more complexity in key 
> generation code.
> 
> Petr Špaček  @  CZ.NIC
> 
> 
> On 03. 07. 20 15:08, libor.peltan wrote:
>> Hi Thomas,
>>
>> well, the key tag is a "checksum" of the crypto key material, which is 
>> "random", so with a probability of 1/65536, you obtain two different keys 
>> with equal key tag.
>>
>> This should be no problem for the operation, you just need to take care and 
>> not use key tag for identifying the keys, for example in keymgr.
>>
>> Note that in Knot terminology, this short checksum is called 'key tag', 
>> whereas 'key id' is a sha sum which is most probably unique.
>>
>> BR,
>>
>> Libor
>>
>> Dne 03.07.20 v 14:58 Thomas E. napsal(a):
>>> Hi,
>>>
>>> I performed a manual key roll over with this command:
>>>
>>> $ knotc  zone-key-rollover dnssec-test.xxx zsk
>>>
>>>
>>> The result is 2 different ZSK's with the same key id:
>>>
>>>
>>> dnssec-test.xxx.    3600 IN    DNSKEY 256 3 8 (
>>>                 AwEAAc5W.....
>>>                 ) ; ZSK; alg = RSASHA256; key id = 7030
>>>
>>> dnssec-test.xxx.    3600 IN    DNSKEY 256 3 8 (
>>>                 AwEAAc7Q5U......
>>>                 ) ; ZSK; alg = RSASHA256; key id = 7030
>>>
>>>
>>>
>>>  From the log:
>>>
>>> 2020-07-03T14:52:59 info: [dnssec-test.xxx.] DNSSEC, key, tag 56464,
>>> algorithm RSASHA256, KSK, public, ready, active+
>>> 2020-07-03T14:52:59 info: [dnssec-test.xxx.] DNSSEC, key, tag  7030,
>>> algorithm RSASHA256, public
>>> 2020-07-03T14:52:59 info: [dnssec-test.xxx.] DNSSEC, key, tag  7030,
>>> algorithm RSASHA256, public, active
>>> 2020-07-03T14:52:59 info: [dnssec-test.xxx.] DNSSEC, signing started
>>> 2020-07-03T14:52:59 info: [dnssec-test.xxx.] DNSSEC, zone is up-to-date
>>>
>>>
>>>
>>> Is it the indented behavior to have two ZSK's with the same key id?
>>>
>>>
>>> Thanks a lot,
>>> Thomas
> 

-- 
https://lists.nic.cz/mailman/listinfo/knot-dns-users

Reply via email to