It was reported  that when the BIND authoritative name server daemon (named) processed a successful IXFR transfer or a dynamic update, there was a small period of time during which the IXFR/update, along with a query, could cause the server to stop processing all requests. A higher update and/or query rate would increase the probability of the deadlock occurring.
This flaw only affects BIND 9.7.1 and 9.7.2; upstream has released 9.7.3 to
correct this flaw. Upstream also documents that using the "-n1" option to
cause named to use only one worker thread would mitigate this problem.
Upstream verified to me that this was introduced in 9.7.1, so bind in RHEL6 is not vulnerable.
The fix is also noted as:
Corrected a defect where a combination of dynamic updates and zone transfers incorrectly locked the in-memory zone database, causing named to freeze. [RT #22614]
9.7.3 is currently in Fedora 13 and 14 testing repositories.
Not vulnerable. This issue did not affect the versions of bind as shipped with Red Hat Enterprise Linux 4, 5, or 6.