Round Robin fails in bind 9.2.1 for a large packet (TCP) or big number of items

Manuel manfer15 at terra.es
Fri Nov 28 11:39:07 UTC 2003


I have a bind 9.2.1 installation to support Microsoft Active
Directory. The case is that in one of my Microsoft domains I have
about 40 domain controlers. then when a windows 2000/XP client ask for
logon to domain always ask the same Domain controler for logon and
site asignation, that is the first resource record returned by BIND in
a respose to a question for resolve the A records for "mydomain.com"
that is because BIND servers never Round Robin for this question and
always return the A registers in the same order.
I have anothers Microsoft domains and the Round Robin works fine,
because I have only 2 or 6 Domain Controlers for these Domains and the
Round Robin always works.
I´m looking for solutions in bind 9 but the only thing I see is
rrset-order option but it is not yet implemented in bind 9.2.1


Is there any patch for BIND to solve this bug?
Is there anybody with the same problem? and if this is the case, Some
sugestions to solve this matter?


Thanks a lot
Regard
Manuel Fernández


More information about the bind-users mailing list