For lack of a better title. Our environment requires having subdomains where we allow dynamic updates using bind as the DNS server and its nsupdate utility authenticating via keyfiles. We're looking to make this more redundant. Currently we have one primary name server for all of these subdomains and several secondaries for our top level domain, (but not for the subdomains). As an example using bar.com as the TLD and foo.bar.com as the subdomain. The foo.bar.com subdomain zone file would have entries similar to:
NS ns1.bar.com
A 1.2.3.4
MX mx1.bar.com
www A 1.2.3.5
...
and the primary zone file would have entries for 'foo' such as:
foo NS ns1.bar.com
Currently the subdomains are not replicated to other name servers, i.e. ns1 is the only name server that can resolve these even though ns2,ns3,etc. are setup as secondaries for the bar.com domain. Ideas on how to set this up in a more reliable way? If the glue record in the bar.com zone file was removed and multiple NS records setup in each subdomain (and set them up as slaves on secondary name servers) that would help resolve the domains, but it wouldn't help with providing an alternate 'update' server should a subdomain need to be updated. If ns1 was down in this example, how would one allow the nsupdate clients to update their zone and have it replicate to the other name servers as well as when ns1 comes back online? Is this even possible with Bind?