v$logmnr_contents sql_undo/sql_redo 4000 bytes limit
604207Nov 17 2009 — edited Dec 29 2009There is a limitation on the SQL_UNDO/SQL_REDO columns in v$logmnr_contents view, whenever the redo or undo value exceeds this 4000 bytes limit the records will be split.
There is an option to track the split records using the RS_ID and CSF(Continuation SQL Flag).
CSF - 1 indicating the that either SQL_REDO or SQL_UNDO is greater than 4000 bytes in size and is continued in the next row returned by the view,
- 0 indicating SQL_REDO and SQL_UNDO is contained within the same row.
However we will be able to identify the split records in v$logmnr_contents view alone, the moment we store the data into any physical table. There is no guarantee that the split records will be returned in the same order. So there is a chance that we will not be able to reconstruct the exact sql_undo and sql_redo statements.
Oracle doesn't seem to handle this, is there a way to achieve this.