I have run bind name servers for large systems for a long time. My experience is that these events are very often caused by serial number problems in the S\Start of Authority (SOA) record. This may not be the problem here, but remember when you change a DNS master file, for the changes to propagate its serial number must increase. A slave (secondary) server will not accept a new version of the file unless the new serial number is greater than the one in the version it currently holds. I used a Perl script that caused serial numbers to be generated with each change in the form: YYYYMMDDNN. (Year, Month & Day of the change, and the NN allowed multiple changes per DAY. If you have a seriously disrupted serial number. Setting the serial number to "0" and propagating it will allow you to get the slaves back in sync. I hope this helps. dlg David L. Gehrt 1865 Wilding Lane San Luis Obispo, CA 93401-3044 Email: dlg@xxxxxxxxxxx
