Archives of the TeradataForum

Message Posted: Thu, 09 Dec 2004 @ 11:18:45 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

Hello,

Clay Barrineau wrote:

Though the chances of 2641 happening are smaller, LOCKING FOR ACCESS will not guarantee it won't happen. Because, when I explain CREATE JOIN INDEX (and CREATE INDEX), it shows, that the table is locked for exclusive use at some point in time - late in the creation of the index.

Anyway, a request for a read lock on a table should not receive an error, if an index has been added to that table. I can understand the need for this if the request required a write lock - then maybe the generated execution plan could cause inconsistencies. In my opinion, Teradata should be able to do better in this case. It would be nice, if someone from Teradata would comment...

Anyway, if BTEQ is used, and RETRY IS ON, then the command ".SET ERRORLEVEL 2641 SEVERITY 0" would solve the problem. And this solution is not dangerous by any means, because what would happen is the following: The request getting the 2641-error would automatically be resubmitted (and parsed again), which is exactly what Teradata wants (2641 error text is: "%DBID.%TVMID was restructured. Resubmit.").

Best Regards,

Ole Dunweber
Coop Nordic

	<Prev	Next>		<<First	<Prev	Next>	Last>>