Description of the reasons that the SQL Server transaction log is not being truncated (62866)
The information in this article applies to:
- Microsoft SQL Server 4.2x
- Microsoft SQL Server 6.0
- Microsoft SQL Server 6.5
- Microsoft SQL Server 7.0
- Microsoft SQL Server 2000 (all editions)
- Microsoft SQL Server 2005 Standard Edition
- Microsoft SQL Server 2005 Developer Edition
- Microsoft SQL Server 2005 Enterprise Edition
- Microsoft SQL Server 2005 Express Edition
- Microsoft SQL Server 2005 Workgroup
This article was previously published under Q62866 SUMMARY
Listed below are some reasons why the transaction log fills up and never gets truncated, even though the "trunc. log on chkpnt" option is set to true and regular checkpoints are done during processing.
MORE INFORMATION
An unbounded update fills up the log; that is, there is no WHERE clause on
the UPDATE statement. Since a transaction (either user declared or
implicit) must be able to be entirely rolled back or committed as a whole,
the log must be large enough to maintain all information for the
transaction. Thus, even when truncating the log, it is possible to fill the
log if it is too small for a single large transaction.
This situation can also occur if a WHERE clause is used, the table is very
large, and the WHERE clause is too general, causing a large number of
records to be affected. There is another case where this could happen: if
multiple updates are performed that together affect enough records to fill
the log before the checkpoint is done, the log can still fill up.
| Modification Type: | Major | Last Reviewed: | 12/28/2005 |
|---|
| Keywords: | kbinfo kbother KB62866 |
|---|
|