Description of Problem:
RedHat Network has bind-9.2.1-0.70.2 in place to update earlier versions to fix
a resolver problem. However, in updating a slave name server to this version,
zone transfers now fail quickly with "refresh: failure trying master <server
info>: time out".
This was a working, functional secondary name server for many hundreds of
zones. We're now trying to figure out how to get back to Bind-8 so that we can
return this server to functional service.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
From the same server where BIND claims AXFRs are timing out, I am able to do:
dig @masterserver zonename axfr
and get a nice, complete zone transfer to my screen.
Clearly, the master server is having no trouble serving up the zone transfers.
The problem appears confined to the slave server.
Another update on this issue. Since Bind-9 uses separate configuration items
for transfer sorce and notify source, those items affected the operation of
the servers in question.
This points out the folly of using a new release of a software product as
an "errata" fix, without substantial documentation or warning. Errata should
NOT break running systems.
Going from a bind-8 to a bind-9 to fix a bug may well have been the shortest
path, but may well have cost many customers a lot of money figuring out and
fixing systems which became inoperative. It's also possible this upgrade
opened customers up to new vulnerabilities, as BIND-9 uses encryption
libraries which have been the subject of errata.
This was an upgrade issue, that should have been documented better.