BUG: INSERT..SELECT with Ignore_Dup_Row Clustered Index Stops

• Microsoft SQL Server, version 6.5

SYMPTOMS

INSERT..SELECT with ignore_dup_row on a clustered index stops responding, and the insert never completes. Examination of sysprocesses shows the spid to be sleeping, with a waittype of 0x0013. No indication of blocking nor deadlocking is present; effectively, the insert has blocked itself.

WORKAROUND

To work around this problem, replace the clustered index with a non-clustered index. Alternatively, removing the ignore_dup_key option from the clustered index may solve the problem.

STATUS

Microsoft has confirmed this to be a problem in SQL Server version 6.5. We are researching this problem and will post new information here in the Microsoft Knowledge Base as it becomes available.

MORE INFORMATION

The spid will not respond to a KILL statement; SQL Server must be shut down. Depending of the size of the insert, there may be a lengthy recovery process for the database when SQL Server is restarted.