Tracing and Instrumenting Applications

บทความ
09/15/2021

Note

This article is specific to .NET Framework. It doesn't apply to newer implementations of .NET, including .NET 6 and later versions.

Tracing is a way for you to monitor the execution of your application while it is running. You can add tracing and debugging instrumentation to your .NET Framework application when you develop it, and you can use that instrumentation both while you are developing the application and after you have deployed it. You can use the System.Diagnostics.Trace, System.Diagnostics.Debug, and System.Diagnostics.TraceSource classes to record information about errors and application execution in logs, text files, or other devices for later analysis.

The term instrumentation refers to an ability to monitor or measure the level of a product's performance and to diagnose errors. In programming, this means the ability of an application to incorporate:

Code tracing - Receiving informative messages about the execution of an application at run time.
Debugging - Tracking down and fixing programming errors in an application under development. For more information, see Debugging.
Performance counters - Components that allow you to track the performance of your application. For more information, see Performance Counters.
Event logs - Components that allow you to receive and track major events in the execution of your application. For more information, see the EventLog class.

Instrumenting your application by placing trace statements at strategic locations in your code is especially useful for distributed applications. By using trace statements you can instrument an application not only to display information when things go wrong, but also to monitor how well the application is performing.

The TraceSource class provides enhanced tracing features and can be used in place of the static methods of the older Trace and Debug tracing classes. The familiar Trace and Debug classes are still widely used, but the TraceSource class is recommended for new tracing commands, such as TraceEvent and TraceData.

The Trace and Debug classes are identical, except that procedures and functions of the Trace class are compiled by default into release builds, but those of the Debug class are not.

The Trace and Debug classes provide the means to monitor and examine application performance either during development or after deployment. For example, you can use the Trace class to track particular types of actions in a deployed application as they occur (for example, creation of new database connections), and can therefore monitor the application's efficiency.

Code Tracing and Debugging

During development, you can use the output methods of the Debug class to display messages in the Output window of the Visual Studio integrated development environment (IDE). For example:

Trace.WriteLine("Hello World!")
Debug.WriteLine("Hello World!")

System.Diagnostics.Trace.WriteLine("Hello World!");
System.Diagnostics.Debug.WriteLine("Hello World!");

Each of these examples will display "Hello World!" in the Output window when the application is run in the debugger.

This enables you to debug your applications and optimize their performance based on their behavior in your test environment. You can debug your application in your debug build with the Debug conditional attribute turned on so that you receive all debugging output. When your application is ready for release, you can compile your release build without turning on the Debug conditional attribute, so that the compiler will not include your debugging code in the final executable. For more information, see How to: Compile Conditionally with Trace and Debug. For more information on different build configurations for your application, see Compiling and Building.

You can also trace code execution in an installed application, using methods of the Trace class. By placing Trace Switches in your code, you can control whether tracing occurs and how extensive it is. This lets you monitor the status of your application in a production environment. This is especially important in a business application that uses multiple components running on multiple computers. You can control how the switches are used after deployment through the configuration file. For more information, see How to: Create, Initialize and Configure Trace Switches.

When you are developing an application for which you intend to use tracing, you usually include both tracing and debugging messages in the application code. When you are ready to deploy the application, you can compile your release build without turning on the Debug conditional attribute. However, you can turn on the Trace conditional attribute so that the compiler includes your trace code in the executable. For more information, see How to: Compile Conditionally with Trace and Debug.

Phases of Code Tracing

There are three phases of code tracing:

Instrumentation — you add trace code to your application.
Tracing — the tracing code writes information to the specified target.
Analysis — you evaluate the tracing information to identify and understand problems in the application.

During development, all debug and trace output methods write information to the Output window in Visual Studio by default. In a deployed application, the methods write tracing information to the targets you specify. For more information on specifying an output target for tracing or debugging, see Trace Listeners.

The following is an overall view of the major steps typically involved in using tracing to analyze and correct potential problems in deployed applications. For more information on how to perform these steps, see the appropriate link.

To use tracing in an application

Consider which tracing output you will want to receive onsite after you have deployed the application.
Create a set of switches. For more information, see How to: Configure Trace Switches.
Add the trace statements to the application code.
Determine where you want the tracing output to appear and add the appropriate listeners. For more information, see Creating and Initializing Trace Listeners.
Test and debug your application and the tracing code it contains.
Compile the application into executable code using one of the following procedures:
- Use the Build menu along with the Debug page of the Property Pages dialog box in Solution Explorer. Use this when compiling in Visual Studio.
  
  - or -
- Use the Trace and Debug compiler directives for the command-line method of compiling. For more information, see Compiling Conditionally with Trace and Debug. Use this when compiling from the command line.
If a problem occurs during run time, turn on the appropriate trace switch. For more information, see Configuring Trace Switches.

The tracing code writes tracing messages to a specified target, for example, a screen, a text file, or an event log. The type of listener you included in the Trace.Listeners collection determines the target.
Analyze the tracing messages to identify and understand the problem in the application.

Trace Instrumentation and Distributed Applications

When you create a distributed application, you might find it difficult to test the application in the manner in which it will be used. Few development teams have the capability to test all possible combinations of operating systems or Web browsers (including all the localized language options), or to simulate the high number of users that will access the application at the same time. Under these circumstances, you cannot test how a distributed application will respond to high volumes, different setups, and unique end-user behaviors. Also, many parts of a distributed application have no user interface with which you can interact directly or view the activity of those parts.

However, you can compensate for this by enabling distributed applications to describe certain events of interest to system administrators, especially things that go wrong, by instrumenting the application — that is, by placing trace statements at strategic locations in your code. Then if something unexpected occurs at run time (for example, excessively slow response time), you can determine the likely cause.

With trace statements you can avoid the difficult task of examining the original source code, modifying it, recompiling, and attempting to produce the run-time error within the debugging environment. Remember that you can instrument an application not only to display errors, but also to monitor performance.

Strategic Placement of Trace Statements

You must exercise special care when placing your trace statements for use during run time. You must consider what tracing information is likely to be needed in a deployed application, so that all likely tracing scenarios are adequately covered. Because applications that use tracing vary widely, however, there are no general guidelines for strategic placement of tracing. For more information on placing trace statements, see How to: Add Trace Statements to Application Code.

Output from Tracing

Trace output is collected by objects called listeners. A listener is an object that receives trace output and writes it to an output device (usually a window, log, or text file). When a trace listener is created, it is typically added to the Trace.Listeners collection, allowing the listener to receive all trace output.

Tracing information is always written at least to the default Trace output target, the DefaultTraceListener. If for some reason you have deleted the DefaultTraceListener without adding any other listeners to the Listeners collection, you will not receive any tracing messages. For more information, see Trace Listeners.

The six Debug members and Trace methods that write tracing information are listed in the following table.

Method	Output
`Assert`	The specified text; or, if none is specified, the Call Stack. The output is written only if the condition specified as an argument in the `Assert` statement is false.
`Fail`	The specified text; or, if none is specified, the Call Stack.
`Write`	The specified text.
`WriteIf`	The specified text, if the condition specified as an argument in the `WriteIf` statement is satisfied.
`WriteLine`	The specified text and a carriage return.
`WriteLineIf`	The specified text and a carriage return, if the condition specified as an argument in the `WriteLineIf` statement is satisfied.

All listeners in the Listeners collection receive the messages described in the above table, but the actions taken may vary depending on what kind of listener receives the message. For example, the DefaultTraceListener displays an assertion dialog box when it receives a Fail or failed Assert notification, but a TextWriterTraceListener simply writes the output to its stream.

You can produce custom results by implementing your own listener. A custom trace listener might, for example, display the messages to a message box, or connect to a database to add messages to a table. All custom listeners should support the six methods mentioned above. For more information on creating developer-defined listeners, see TraceListener in the .NET Framework reference.

The Write and WriteLine methods always write the text that you specify. Assert, WriteIf, and WriteLineIf require a Boolean argument that controls whether or not they write the specified text; they write the specified text only if the expression is true (for WriteIf and WriteLineIf), or false (for Assert). The Fail method always writes the specified text. For more information, see How to: Add Trace Statements to Application Code and the .NET Framework reference.

Security Concerns

If you do not disable tracing and debugging before deploying an ASP.NET application, your application may reveal information about itself that could be exploited by a malicious program. For more information, see How to: Compile Conditionally with Trace and Debug, Compiling and Building, and How to: Create, Initialize and Configure Trace Switches. Debugging is also configurable through Internet Information Services (IIS).

แชร์ผ่าน