Slave's lack of NS records cause zone file dump
bind-users at dollardns.net
Fri Oct 22 15:58:11 UTC 2004
> There is nothing in the protocols that says a name server must be listed
> in the NS RR set in order to sucessfully complete a zone transfere and
> slave the zone. However, a name server slaving a zone in this fashion
> would probably see little if any requests for information in that zone
> from other name servers, as there is no information in the name space to
> tell people that the name server in question is actually slaving the zone.
This wasn't really the point. The point is that it starts out working and stops working in the event of a restart. Either it should always work or never work. Mark chose the never work option since a zone should always have authority records. He's escalated this to ISC-Bugs #12894 and provided a patch.
> Complains how? In the log? Could you share the log?
Yes, it complains in the general logging category. You get the normal dns_from_master or whatever it is called saying there's a lack of NS records when loading the zone. Mark referenced the exact error message in his reply to my bug report. If you want to see the message, then try to load a zone without NS records.
> am not sure what is causing the crash, but, I am fairly certain that it
> is not a lack of NS RR's in a master zone db file on a different name
Actually I have a good handle on what's going on. The problem exists exactly for the reason I said it exists. The result is exactly as I stated it. If it was necessary I could've provided further proof of concept, but it appears Mark understood me completely.
More information about the bind-users