beginObjectTracking(fileName: String);
The beginObjectTracking method of the System class starts recording, in a file (on the database server node) specified by the fileName parameter, every persistent object read operation from the database or write operation to the database. The information recorded distinguishes between read operations to get an object and read operations to lock an object. For write operations, it also distinguishes between write operations to create an object, write operations to delete an object, and write operations to update an object.
This does not necessarily record every time an application uses an object, because if the object resides in the persistent object cache, it may not have to be fetched from the database.
Object tracking is terminated by using the endObjectTracking method defined in the System class.
The file is a standard text file. The first line is a header record and the last line is a trailer record. The lines in between represent individual object accesses. In each line, fields are separated by spaces.
The format of the three types of records is similar to that used for node sampling files. The format of the header record is shown in the following table.
For details, see "
Only one object tracking session can be active at a time. If the beginObjectTracking method is called when object tracking is already active, an 1138 exception (
Use this method with caution. When object tracking is active, the tracking file can fill very rapidly. Object tracking should therefore be used in relatively short bursts.