[CR #400988] If a server crashes following an alter table unpartition operation of very large tables, the server recovery after restart can leave the table in an inconsistent or corrupt state. dbcc check table and other commands affecting table’s pages can result in wrong page errors. In some cases, the recovery process encounters timestamp mismatches on pages allocated to the table being unpartitioned. This form of corruption is possible when you issue a shutdown no wait command, and affect very large tables with multiple OAM pages in the OAM chain for each partition.
Workaround: Run "CHECKPOINT
<database>" immediately following an unpartition operation.
In addition, a polite shutdown avoids this form
of corruption.
| Copyright © 2005. Sybase Inc. All rights reserved. |
|
|
