![]() |
![]() HP OpenVMS Systemsask the wizard |
![]() |
The Question is: I'm currently experimenting with the BIND 8.1-2 server included with TCIP/IP V5.0 on AXP OpenVMS 7.2. I've loaded the zone file for our domain (about 8300 RRs). Queries for individual hosts work correctly. I have not tried a zone transfer between master and a secondary server yet. When using nslookup, if I try to "ls" the domain the server I sometimes get the following error after several screenfulls scroll by: *** ls: error receiving zone transfer: result: NOERROR, answers = 1, authority = 0, additional = 0 *** Can't list domain domain-name-here: Unspecified error Where it fails is variable. Sometimes early, sometimes late. Sometimes it works through to completion without error, but not often. The exact same message occurs when running nslookup from other systems against the same VMS-based server. For example, DEC Unix 4.0e. The command TCPIP SHOW HOST produces a similar error (again after receiving many screenfulls): $ TCPIP SHOW HOST %NONAME-E-NOMSG, Message number 00000002 %TCPIP-W-NORECORD, information not found -NONAME-E-NOMSG, Message number 00000002 As a bit of an aside, a cohort built the new ISC BIND 8.2 distribution on the 4.0e system and loaded the exact same zone and conf files. The 8.2 BIND server works perfectly, using either the VMS TCP/IP nslookup or the Unix system's nslookup. Any ideas? BIND 8.2 has a number of security fixes in it over the BIND 8.1-x release. When can we expect the new version to be incorporated into the TCP/IP product? Thanks, Cullen The Answer is : This problem has been previously reported and a fix has been created. The fix will be available as part of the TCP/IP Services V5.0A update. The underlying problem permitted a send buffer to be allocated that was too large to be properly handled by a write() operation. Due to scheduling issues, the next release of BIND will likely be based on V9, and not BIND 8.2.
|