On Nov 23, 2017, at 06:19, Havard Eidnes <h...@uninett.no> wrote: > Secondly: can someone please explain to me why the idea of a > "primary master" where the zone data originates from and where > updates are performed is considered archaic?
I think the only in-protocol use of the MNAME field is to specify the name to which UPDATE messages are sent. The MNAME was originally used to specify the name of a single master server to which slaves would send zone transfer requests, and from which zone data would be served in response, but as far as I am aware this was only ever informational: that is, I'm not aware of any implementation of a slave server that automatically detected its master by looking for the MNAME. Today I think it's fair to say that any non-trivial DNS deployment that uses zone transfers makes use of a non-trivial graph for distribution of zone transfers, with slaves configured to send zone transfer requests to more than one master server, and in many cases with layers of servers that act as slaves and masters simultaneously for reasons of redundancy and scaling. While a single slave sending zone transfer requests to a single master is still a valid example of such a graph, I think it's more a degenerate case than the usual case today. In that sense the idea of using a single master (which I think is implied by "primary master" and a name published in a single MNAME field) is defensibly archaic. My perspective may not be universal, of course, but that's what I see. Joe _______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop
