Redo and Create Table As Select ...
441391Apr 2 2005 — edited Apr 5 2005Since a Create table statement is DDL is the whole statement written to redo?
I have a situation where I am testing an unrecoverable operation creating a table as a select statement. I have three basic test cases:
1. Execute CTAS, commit, and then shutdown abort. After startup, the table is there with the data. (As expected)
2. Execute CTAS, do NOT commit, and then shutdown abort. After startup, the table is there with the data. (NOT expected)
3. Execute CTAS and then shutdown abort in the middle of the transaction. The table is not there. (as expected, although we did expect corrupt segments of the tablespace, and thus expected having to take it offline, but this didn't happen.)
Is the whole statement being sent to redo, and being executed on startup in the 2nd scenario? Is there an explicit commit in a CTAS? (I am making sure autocommit is off)
Thanks for any help!