A typical session involves:

1. (Optional) Limit tracing to specific subprograms and choose a tracing level.

   Tracing all subprograms and exceptions in a large program can produce huge amounts of data that are difficult to manage.

2. Starting PL/SQL tracing in session (DBMS_TRACE.SET_PLSQL_TRACE).

3. Running an application to be traced.

4. Stopping PL/SQL tracing in session (DBMS_TRACE.CLEAR_PLSQL_TRACE).

After you have collected data with Trace, you can query the database tables that contain the performance data and analyze it in the same way that you analyze the performance data from Profiler.

These constants are shown in the following table.

Oracle recommends using the symbolic form for all these constants.

These are described in the following sections:

• Controlling Data Volume

• Creating Database Tables to Collect DBMS_TRACE Output

• Collecting Trace Data

• Collected Data

• Trace Control

Controlling Data Volume

Profiling large applications may produce a large volume of data. You can control the volume of data collected by enabling specific program units for trace data collection.

You can enable a program unit by compiling it debug. This can be done in one of two ways:

alter session set plsql_debug=true; 
create or replace ... /* create the library units - debug information will be generated */ 

or:

/* recompile specific library unit with debug option */ 
alter [PROCEDURE | FUNCTION | PACKAGE BODY]  <libunit-name> compile debug; 

You can limit the amount of storage used in the database by retaining only the most recent 8,192 records (approximately) by including TRACE_LIMIT in the TRACE_LEVEL parameter of the SET_PLSQL_TRACE procedure.

Creating Database Tables to Collect DBMS_TRACE Output

You must create database tables into which the DBMS_TRACE package writes output. Otherwise, the data is not collected. To create these tables, run the script TRACETAB.SQL. The tables this script creates are owned by SYS.

Collecting Trace Data

The PL/SQL features you can trace are described in the script DBMSPBT.SQL. Some of the key tracing features are:

• Tracing Calls

• Tracing Exceptions

• Tracing SQL

• Tracing Lines

Additional features of DBMS_TRACE also allow pausing and resuming trace, and limiting the output.

Tracing Calls

Two levels of call tracing are available:

• Level 1: Trace all calls. This corresponds to the constant TRACE_ALL_CALLS.

• Level 2: Trace calls to enabled program units only. This corresponds to the constant TRACE_ENABLED_CALLS.

Enabling cannot be detected for remote procedure calls (RPCs); hence, RPCs are only traced with level 1.

Tracing Exceptions

Two levels of exception tracing are available:

• Level 1: Trace all exceptions. This corresponds to TRACE_ALL_EXCEPTIONS.

• Level 2: Trace exceptions raised in enabled program units only. This corresponds to TRACE_ENABLED_EXCEPTIONS.

Tracing SQL

Two levels of SQL tracing are available:

• Level 1: Trace all SQL. This corresponds to the constant TRACE_ALL_SQL.

• Level 2: Trace SQL in enabled program units only. This corresponds to the constant TRACE_ENABLED_SQL.

Tracing Lines

Two levels of line tracing are available:

• Level 1: Trace all lines. This corresponds to the constant TRACE_ALL_LINES.

• Level 2: Trace lines in enabled program units only. This corresponds to the constant TRACE_ENABLED_LINES.

When tracing lines, Oracle adds a record to the database each time the line number changes. This includes line number changes due to procedure calls and returns.

Collected Data

If tracing is requested only for enabled program units, and if the current program unit is not enabled, then no trace data is written.

When tracing calls, both the call and return are traced. The check for whether tracing is "enabled" passes if either the called routine or the calling routine is "enabled".

Call tracing will always output the program unit type, program unit name, and line number for both the caller and the callee. It will output the caller's stack depth. If the caller is enabled, the caller's name will also be output. If the callee is enabled, the callee's name will also be output.

Exception tracing writes out the line number. Raising the exception shows information on whether the exception is user-defined or pre-defined. It also shows the exception number in the case of pre-defined exceptions. Both the place where the exceptions are raised and their handler is traced. The check for tracing being "enabled" is done independently for the place where the exception is raised and the place where the exception is handled. Enabling NO_TRACE_HANDLED_EXCEPTIONS limits data collection to unhandled exceptions

All calls to DBMS_TRACE.SET_PLSQL_TRACE and DBMS_TRACE.CLEAR_PLSQL_TRACE place a special trace record in the database. Therefore, it is always possible to determine when trace settings were changed.

Trace Control

As well as determining which items are collected, you can pause and resume the trace process. No information is gathered between the time that tracing is paused and the time that it is resumed. The constants TRACE_PAUSE and TRACE_RESUME are used to accomplish this. Trace records are generated to indicate that the trace was paused/resumed.

It is also possible to retain only the last 8,192 trace events of a run by using the constant TRACE_LIMIT. This allows tracing to be turned on without filling up the database. When tracing stops, the last 8,192 records are saved. The limit is approximate, since it is not checked on every trace record. At least the requested number of trace records will be generated; up to 1,000 additional records may be generated. At least the requested number of trace records will be generated; up to 1,000 additional records may be generated. The 8,192 record limit can be changed. Setting event 10940 to level n changes the record limit to 1024 * n.

Enabling NO_TRACE_ADMINISTRATIVE prevents the generation of such administrative event records as PL/SQL Trace Tool started, Trace flags changed, PL/SQL Virtual Machine started, and PL/SQL Virtual Machine stopped.