What's puzzling is the developer is claiming that no ORA- errors were seen, so even with a PK or unique key violation that should throw an error.
Does Oracle actually apply the transaction then roll it back in cases of PK/Unique viloations?
Interesting, I'll test for that thanks! Ahbaid Finn Jorgensen wrote:
Barring a user rollback operation, perhaps an insert into a table with a PK or unique index which was violated? FinnOn Dec 8, 2007 2:59 PM, Ahbaid Gaffoor <ahbaid@xxxxxxx <mailto:ahbaid@xxxxxxx>> wrote:We're trying to figure out exactly what an application has done in the db. I restored an archive log from tape and mined it. When querying the logminer view I'm seeing two rows for a transaction in V$LOGMNR_CONTENTS The first row has sql entries in SQL_REDO and SQL_UNDO, and the ROLLBACK field is 0 The second row has sql only in SQL_REDO and the ROLLBACK field is 1 Both rows have the same SCN and TIMESTAMP, does this mean that the first row was applied and the second was indeed a rollback entry? What's especially puzzling is we are not seeing any lost connections or abnormal program terminations, so what's causing this rollback entry? all ideas and comments welcome. thanks Ahbaid -- //www.freelists.org/webpage/oracle-l
