Hi,
indeed, the TTLs of a single RRSet are required to be the same according
to the RFC mentioned below. However, we are aware that failing to load
the zone because of this may not be desirable, so the upcoming release
of Knot will give a notice of the TTL mismatch, but set the TTLs of all
RRs in a RRSet to a common value (that of the last RR).
Mark Karpilovskij
On 31.7.2018 09:38, Petr Špaček wrote:
On 31.7.2018 08:53, Zdenek Novy wrote:
Hi,
I would like to ask about the implementation of the Resource records in
RRSet in KnotDNS.
I have the domain with the three TXT record with same class IN for the
same label ('@') and with the different TTLs. In nsd and bind DNS
servers seems everything fine, but in KnotDNS I got the warning and error:
knotd[551]: warning: [xxxxxxx.xxx.] zone loader, RRSet TTLs mismatched,
node 'xxxxxxx.xxx.' (record type TXT)
knotd[551]: error: [xxxxxxx.xxx.] zone loader, failed to load zone, file
'/etc/knot/files/master.gen/xxxxxxx.xxx' (TTL mismatch)
knotd[551]: error: [xxxxxxx.xxx.] failed to parse zonefile (failed)
knotd[551]: error: [xxxxxxx.xxx.] zone event 'load' failed (failed)
Is it a correct behavior and other DNS servers don't check it or is it a
bug in KnotDNS?
This behavior is correct according to standard
https://tools.ietf.org/html/rfc2181#section-5.2 :
Consequently the use of differing TTLs in an RRSet is hereby
deprecated, the TTLs of all RRs in an RRSet must be the same.
I suspect that other servers will silently clip TTL to one value but I
haven't checked that.
Petr Špaček @ CZ.NIC
> Thank you for reply.
> Cheers,