Tutorial

Logging & Tracing

Mechanism in
SAP
Version 1.2

Infrastructure Management Team

Table of Contents
1.

ABOUT THIS TUTORIAL ............................................................................................................................ 5

1.1
1.2
1.3

2.

TARGET AUDIENCE .................................................................................................................................... 5

THINGS YOU SHOULD KNOW IN ADVANCE .................................................................................................. 5
DOWNLOAD & JAVADOC ................................................................... ERROR! BOOKMARK NOT DEFINED.

INTRODUCTION ........................................................................................................................................... 6
2.1
WHAT IS LOGGING AND TRACING? ............................................................................................................ 6
2.1.1
Purpose.............................................................................................................................................. 6
2.1.2
Advantage .......................................................................................................................................... 6
2.2
CURRENT PACKAGE ................................................................................................................................... 7
2.2.1
Overall Logic..................................................................................................................................... 7
2.2.2
Conceptual view and terminology ..................................................................................................... 7

3.

SAP LOGGING API BASICS........................................................................................................................ 9

3.1
BACKGROUND ............................................................................................................................................ 9
3.2
COMMON LOGGING METHODS .................................................................................................................... 9
3.3
SIMPLE EXAMPLE FLOW ........................................................................................................................... 10
3.3.1
Tracing ............................................................................................................................................ 10
3.3.2
Explanation...................................................................................................................................... 12
3.3.3
Output .............................................................................................................................................. 12
3.4
CODING RECOMMENDATIONS .................................................................................................................. 13

4.

SAP LOGGING API (DETAILED EXPLANATION) .............................................................................. 14

4.1
DEFAULT SETTING ................................................................................................................................... 14
4.2
DETAILED EXAMPLE ................................................................................................................................ 14
4.2.1
Identify the output source: Location or Category............................................................................ 15
4.2.2
Assign severity to source ................................................................................................................. 16
4.2.3
Specify output destination................................................................................................................ 16
4.2.4
Enable output messages................................................................................................................... 17
4.3
TYPICAL PRACTICE .................................................................................................................................. 22
4.4
BASIC FEATURES ...................................................................................................................................... 22
4.4.1
More configuration options ............................................................................................................. 23
4.4.2
Hierarchical severity inheritance .................................................................................................... 25
4.4.3
Hierarchical destination inheritance............................................................................................... 27
4.4.4
Severity evaluation .......................................................................................................................... 28
4.4.5
Output File....................................................................................................................................... 28
4.4.6
Change Output Logs/Filters ............................................................................................................ 30
4.4.7
Language Dependency..................................................................................................................... 30
4.5
ADVANCED FEATURES ............................................................................................................................. 32
4.5.1
Three forms of log assignment......................................................................................................... 32
4.5.2
Relations between Category & Location ......................................................................................... 34
4.5.3
Relative severities ............................................................................................................................ 35
4.5.4
Multi-threading................................................................................................................................ 39
4.6
CONFIGURATION TOOL ............................................................................................................................. 39
4.6.1
To use............................................................................................................................................... 40
4.6.2
Syntax and semantics of properties file ........................................................................................... 40

5.

ADMINISTRATIVE ISSUE......................................................................................................................... 46
5.1
5.2
5.3

6.

ERROR HANDLING MECHANISM ................................................................................................................ 46

SWITCHING ON/OFF THE TOOL .................................................................................................................. 47
COMPATIBILITY WITH SUN JSR47........................................................................................................... 47

FUTURE DEVELOPMENT ........................................................................................................................ 48

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

6.1
6.2
6.3

STANDARD DEFINITION AND RULES .......................................................................................................... 48

LOG CONSOLE .......................................................................................................................................... 48
THREAD-SPECIFIC TRACING ...................................................................................................................... 48

APPENDIX A ........................................................................................................................................................ 50
SEVERITY LEVELS ................................................................................................................................................ 50
APPENDIX B......................................................................................................................................................... 51
OVERVIEW OF CLASSES HIERARCHY .................................................................................................................... 51
APPENDIX C ........................................................................................................................................................ 52
DEFAULT BEHAVIOR OF CERTAIN OBJECTS ........................................................................................................... 52

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Table of Figures
Figure 2-1 Conceptual View of SAP Logging tool ................................................................................... 8
Figure B-1 Class diagram of the logging API ......................................................................................... 51
Table 2-1 Terminology of key components .............................................................................................. 8
Table 3-1 Common Output API group 1 .............................................................................................. 10
Table 3-2 Common Output API group 2 .............................................................................................. 10
Table 4-1 Placeholders of TraceFormatter .............................................................................................. 24
Table 4-2 Placeholders of filename ......................................................................................................... 29
Table 4-3 Common Output API with support of translation ................................................................... 31
Table 4-4 Setting relative severity........................................................................................................... 36
Table 4-5 Summary of configurable attributes of some classes .............................................................. 44
Table A-1 Definition of severity levels ................................................................................................... 50
Table C-1 Default attributes values......................................................................................................... 52
Table C-2 Highlights of default behavior................................................................................................ 53

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

1. About this tutorial

This tutorial covers the new logging and tracing framework (referred as logging API or SAP
Logging API hereafter), provided for the Java development projects at SAP. The tutorial focuses
on how to use the logging API. The technical implementation details are not covered. Step-by-step
instructions and examples are provided. The goal is to encourage and enable the readers to readily
start using the logging API after finishing this tutorial. Special features made available in this tool
are also introduced in details.
It is recommended to go over the basic concept of the logging framework, as described in Chapter
2 and 3. But for those developers who want to skip the basics and look at some typical logging
operations right away, you may do so by focusing on Chapter 4.2 and 4.6. Other useful basic and
advanced properties are covered in Chapter 4.4 and 4.5 respectively. A few administrative topics
and the outlook on future development are talked about in Chapter 5 and 6.

1.1

Target Audience
Primarily all SAP Java developers who are required to provide logs in a productive environment
for the customers, or who are simply interested in having traces to assist their own debugging tasks
are the audience.

1.2

Things you should know in advance

Prerequisites? In short: Nothing.
Certainly, we expect the developers, who are responsible to instrument the logging message in
their code, have a common sense about where and what to log intelligently. Other than this,
enabling traces or logs is fairly straightforward.
Only when you attempt to use more advanced features, for example, language-neutral message,
you will need to understand the concept of ResourceBundle in Java. (Please refer to Section 4.4.7
for more details about advanced features)

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

2. Introduction
2.1

What is Logging and Tracing?

During the development phase, it is a common practice of developers to insert logging statement
within the code to produce informative logs at the runtime, whether for troubleshooting or for
analysis purposes.
Common problem
Without a standard logging framework, developers are very likely to use System.out.println
in a sporadic manner or call printStackTrace() in case of an exception. Before executing the
code in a productive system, the cleanup of these embedded lines can be very painstaking.
Apparently, this is not very flexible in terms of controlling the amount of log output, the
destination, and the format of the messages, etc.
Two main types
In our logging API, we would even like to further break down the messages into:

Logging: classical log messages, for distinguished problematic areas

Tracing: classical trace messages, for coding problems

2.1.1 Purpose
There are various infrastructure and application development groups in SAP coding in Java, and
either they have their own (quick & dirty) logging mechanism, or they do not do this at all.
However, the availability and readability of traces or event logs are very important for both the
developers and the potential users, like testing group, support group and operators. Therefore, a
common logging framework must be made available to satisfy the requirements of these groups,
especially in a large and distributed environment. A standardized infrastructure will be essential
and beneficial for both the developers and the consumers.

2.1.2 Advantage
In designing this framework, we have kept the following in mind:
Easy to use API
Enabling logs is not a popular task for developer. So, the API/method calls to do so have to be
simple and intuitive.
Performance
The switching on of the logging mechanism should not degrade the performance of the
application as if running with logging,
Easy to maintain
The log insertion done by developer is totally decoupled from executing the code.
Switching on or controlling the amount of log output is configurable at runtime without
modifying the source code.

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

2.2

Current Package

com.sap.tc.logging referred as SAP Logging API, with all functionality for both tracing
and events logging

2.2.1 Overall Logic

General procedures:
1. Identify the source area you would like to produce trace/log output
2. Assign severity level to the source
3. Specify output destination
4. Insert messages with corresponding severity level
Run the program.
When is the message produced?
Only when the severity of the message is equal or higher than the source, the message will be
produced and sent to the destination.
Main tasks
From a developers point of view, normally, you will simply focus on step 1 and 4, which are,
logging meaningful messages at good points of the execution flow. The other 2 steps are more or
less determined and configured by the operators/end-users eventually. This shows the idea of
decoupling the task of enabling the logs by the developers and the program execution by the users.
Sample idea and output
For example (corresponds to the steps mentioned above):
1. Source area: a class fooClass under an arbitrary package com.saps.fooPackage
2. Severity of the source: Debug

Allow messages with severity Debug level or above to be printed

3. Destination: a file
4. Insert an informational text at the very first line of a method, e.g. fooMethod(), indicating
the entry point to this method, with severity Path (higher than Debug)
The message text will be written to the file with a standard format as shown below:
May 3, 2001 6:54:18 PM com.sap.fooPackage.FooClass.fooMethod [main] Path: Entering method

This corresponds to:

Date and timestamp

Full path method name [thread name] Severity: Message

2.2.2 Conceptual view and terminology

Figure2-1 shows a simplified version of the conceptual view that shows the key entities defined in
the SAP Logging API. Brief introduction of the terms is given in Table2-1.

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Log
Log Manager
Manager
1

application
Message
with severity

Log
Log
Controller
Controller
1

Log
Record

Log
Log

publish

1 severity

severity

Filter

Filter

Formatter

Figure 2-1 Conceptual View of SAP Logging tool

Term

Description

Logging Manager
Log Controller
Log Record
Log
Formatter
Filter

A unique single manager that manages mainly the log controllers

Entity that represents the source area
Structure that holds a message and its relevant data
Represents the destination where the message should be output to
Determines the format of the final message text
Optional means to further screen out messages
Table 2-1 Terminology of key components

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

3. SAP Logging API Basics

3.1

Background
The concept of the SAP logging tool is very similar to the other available tools in the market. It
offers better support for a common practice in logging:
Generates classical trace messages and log messages
There exists APIs that are handy to bridge these two types of messages together for advanced
analysis.
Other advanced features will be covered in the later sections.
Are you outputting messages for logging or for tracing?
If your goal is to pinpoint distinguished problematic logistic areas, then you should focus on log
messages, while if you are more interested in analyzing coding problems, then trace messages is
the one to be used.
Yes, this will be the major task as a developer to enable logging: classify the source area that needs
to be logged/traced, and then insert the output messages in the code. Simple enough.
What are the basic classes to start with?
From the conceptual view diagram Figure 2-1, you should know by now the Java class that
represents the source area: Log Controller. In general, you deal with its two subclasses directly:
Category: generate log messages
Location: generate trace messages
Examples for a Category:
/System/Database, System/Security,
Examples for a Location:
com.sap.tc, com.sap.tc.logging,
(Please refer to Appendix B for a complete class hierarchy diagram defined for this logging API).
Lets start getting familiar with the tool by focusing on one area. Since both API for logging and
tracing are quite similar, it will be very straightforward to use one or the other when you have
learnt one of them. Most examples will be focused on the tracing API.

3.2

Common logging methods

With the Location or Category defined, you are ready to insert output methods in your code to fire
messages whenever necessary. Each of the call is already assigned a severity level and this
provides a clean and easy API for users to use readily.
There are a number of methods available to write messages with different severity. They loosely
fall into two groups, where the first group has intuitive name with severity level indicated. These
methods produce messages with the respective severity incorporated in the method name.
According to the severity scale shown in Appendix A, you can find, for example with Location
class:
Common output API with severity indicated
fatalT(string the_message)

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

errorT(string the_message)
warningT(string the_message)
infoT(string the_message)
pathT(string the_message)
debugT(string the_message)
Table 3-1 Common Output API group 1

The second group also has intuitive names, but without explicit severity level shown in their
names. Table 3-2 below shows the ones that are commonly used:
Common output API
(mostly for program flow)
entering()
exiting()
throwing(Throwable the_exception)
assertion(Boolean the_assertion,
String the_message):

Description
Denote method entry with level Severity.Path. Always
used together with exiting().
Denote method exit with level Severity.Path. Always used
together with entering().
Output the exception content with level Severity.Warning
A trace method to verify your assertion and will throw a
Severity.Error message when the condition is false.

Table 3-2 Common Output API group 2

Refer to the Javadocs, it may first appear that there are overwhelmingly numerous methods, but
they are pretty much overloaded methods with different arguments to enhance flexibility. It is
better to provide more options for different requirements, rather than inadequate APIs. (More
details shown in Section 4.2.4).
The APIs shown above are in their simplest form for clarity. But even with these methods, you are
already in good shape to start playing with logging/tracing.

3.3

Simple Example Flow

1. Identify the source area and get a handle to a source object: Location
Name the id of the source location object as the complete package path of the class, e.g.
com.sap.fooPackage.FooClass

2. (Assign severity level to the location object)

3. (Specify output destination for the location object)
4. Insert messages at points where you want to trace the program flow with desired severity level
The common places are: entering/exiting a method, upon throwing an exception, after
performing certain critical tasks, etc
Decide the severity to be assigned to these trace messages.

3.3.1 Tracing
package com.sap.fooPackage;
public class FooClass{
public void method1(){
Tutorial Logging & Tracing Mechanism in SAP
By: Infrastructure Management Team

A typical program
without tracing
inserted.
1


}
public void method2(){
try{
.
}
catch (IOException e){
..
}
}
public static void main(String[] args){
FooClass test1 = new FooClass();
test1.method1();
test1.method2();
}
}

Now, lets insert traces into the program (original code is grayed out for your convenience).
The API to create a source trace object is simple, as well as the output methods used for writing
traces.
package com.sap.fooPackage;
Sample program showing
the instrumentation of
traces

// (0)
import com.sap.tc.logging.*;
public class FooClass{

// (1) The current java class is defined as a location source

private static Location myLoc =
Location.getLocation("com.sap.fooPackage.FooClass");
// (2) Assign severity, otherwise, nothing will be printed out by default
//
Assign destination, no destination by default

static{
myLoc.setEffectiveSeverity(Severity.PATH);
myLoc.addLog(new ConsoleLog());
}
public void method1(){
// (3) Enable the writing of trace messages

myLoc.entering("method1");

myLoc.warningT(method1, Sample warning message);

myLoc.exiting();
}
public void method2(){
myLoc.entering("method2");
try{
.
}
catch (IOException e){
myLoc.throwing(method2, e);
..
}

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

//e.g. file not found exception

myLoc.exiting();
}
public static void main(String[] args){
FooClass test1 = new FooClass();
test1.method1();
test1.method2();
}
}

3.3.2 Explanation
There are mainly a few additional lines you need to activate the tracing feature. Refer the
numbering below to that in the sample source code.
(0): Import the SAP logging package
(1): Get access to a location object for the current class. The API provided should be intuitive
enough: a class method provided by Location class. Users do not have to work on
constructor, just call Location.getLocation(<name of location object>), and
receive a handle to the unique location
(2): Assign severity level and destination for the location object
This will be the level to be compared against the severity of the trace messages. Messages
with severity lower than this of the location object will not be printed. Refer to Appendix A
to get familiar with the severity level definition.
By default, Severity.NONE will be assigned (that is, there will be no output at all). So, for this
sample coding, an explicit level Severity.PATH is assigned to screen out all the debugging
messages, which have a lower severity.
Also by default, no output destination is assigned (again, no output at all). In the example
above, we direct the output to the console. Refer to Appendix B with class diagram that
shows that we mainly have two types of logs: FileLog, ConsoleLog.
(In Section 4.6, you will learn that it is not necessary to hardcode it here in the source, and
you can configure this setting externally without messing with the code.)
(3): Insert trace messages in your program
The example shows a few APIs that write out trace messages: entering, warningT,
throwing, exiting. Which messages will eventually be printed? We know that all these
messages will successfully pass the level Severity.PATH (as specified in this example in step
(2)).

3.3.3 Output
In the example above, messages will be directed to the console with the (default) format readable
by operators (using TraceFormatter).
The content of the output file for the example above looks like this:
May 3, 2001 6:54:18 PM com.sap.fooPackage.FooClass.method1 [main]
Path: Entering method
May 3, 2001 6:54:18 PM com.sap.fooPackage.FooClass.method1 [main]
Warning: Sample warning message
May 3, 2001 6:54:18 PM com.sap.fooPackage.FooClass.method1 [main]
Path: Exiting method
May 3, 2001 6:54:18 PM com.sap.fooPackage.FooClass.method2 [main]
Tutorial Logging & Tracing Mechanism in SAP
By: Infrastructure Management Team

Path: Entering method

May 3, 2001 6:54:18 PM com.sap.fooPackage.FooClass.method2 [main]
Warning: Throwing java.io.FileNotFoundException:
C:\Not_Exist\zzzzz.log (The system cannot find the path
specified)
May 3, 2001 6:54:18 PM com.sap.fooPackage.FooClass.method2 [main]
Path: Exiting method

3.4

Coding Recommendations
The logic so far is straightforward and the learning curve is not steep, just a few additional lines,
and you can enjoy the decent result in a log file.
Before we wrap up this section, there are certain recommendations for you to improve the tracing
functions.
Static reference
As you may have noticed from the sample coding, the variable that holds the location object
is declared with a static modifier to improve efficient access. Basically, a handle to a location
object can always be obtained by calling Location.getLocation(<id>). However,
making such a call every single time may degrade the performance.

Hierarchical naming convention

It is a good idea to create a static variable for the trace object and it is a useful and easy
approach to name the <name of the location object> as the complete package path of the
class, for both clarity and the inheritance features made available in the logging API.
When naming a location object for your Java class, it is a straightforward approach to use a
hierarchical naming convention, e.g. the fully qualified name. This not only avoids ambiguity
on classes having the same name, but also takes the full advantage of the inheritance feature
(e.g. on severity level and output destination) implemented by the logging framework.
The class com.sap.foo is parent to classes com.sap.foo.classA and
com.sap.foo.classB, and the latter two classes are siblings to each other.
In this case, severity level and logs assigned to parent com.sap.foo will become effective
for its children com.sap.foo.classA and com.sap.foo.classB. Siblings will not affect
each other.
Some developers may argue about the potential typing error of a lengthy correct class name
that will affect the inheritance features. That is true. There exists another API for
Location.getLocation(<java class>) which accepts a class other than a string.
Please refer to the JavaDoc for an API that fits your requirements.

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

4. SAP Logging API (Detailed explanation)

4.1

Default Setting
In the previous section, you have already encountered a few times that a default value has already
been set and you simply have to accept the default behavior mentioned in the example. These
settings can be easily manipulated and changed by you through the API or an external
configuration file (details in Section 4.6).
Advantages
Default values are meant to provide a reasonable logging behavior with minimal configuration
required to be done by the users. For example, once a user has specified to output the message to a
console, it is assumed to be read by operators, and thus, a TraceFormatter (formatting message
that is readily understood by human being) is selected with the predefined message pattern that can
clearly displays the message and its relevant info:
May 3, 2001 6:54:18 PM com.sap.fooPackage.FooClass.fooMethod [main]
Path: Entering method

References
There will be more predefined values appearing in the following sections. To assist you in moving
on smoothly with the tutorial, as well as avoiding ambiguity for your future coding, refer to
Appendix C where we are keeping track of the various default settings and behavior.

4.2

Detailed Example
To recap again, there are mainly 4 steps in enabling logging for your application.
Identify source code area, e.g. a class Node, and represent it with a Location object
Location loc = Location.getLocation("<package>.Node");

Assign severity level to the source

loc.setSeverity(Severity.WARNING);

//default: Severity.NONE

Specify output destination

loc.addLog(new ConsoleLog());

//with default formatter: TraceFormatter

Instrument trace message with specified severity

loc.entering(methodname);
loc.debugT(methodname, message);
loc.fatalT(methodname, message);

It is a legitimate deduction that more than 90% of your development time will be spent on step 1
and 4. Eventually, step 2 and 3 can be mainly configured externally so that users can control the
output without the need to modify source code and recompile again.
Shown below is a very basic sample code, showing step 1 & step 4:
package com.sap.fooPackage;
import com.sap.tc.logging.*;
public class Node {
private static final Location loc =
Location.getLocation("com.sap.fooPackage.Node");
public void announce(Object o) {
String method = "announce(java.lang.Object)";

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

loc.entering(method);
try{
// do something...
loc.debugT(method, "Connecting to ");
}
catch (Exception e) {
loc.fatalT(method,
"Error processing object {0}",
new Object[] {o});
}
loc.exiting();
}
}

Step 2 & 3 are not shown at this point, but assuming the severity level assigned to this is set to be
Severity.ALL (accept all severity levels and output everything) and output has been piped to a
ConsoleLog (terminal).
The output will look like this, formatted with TraceFormatter (default formatter for ConsoleLog).
May 3, 2001 6:54:18 PM com.sap.fooPackage.Node.announce [main] Path: Entering method
May 3, 2001 6:54:18 PM com.sap.fooPackage.Node.announce [main] Debug: Connecting to .
May 3, 2001 6:54:18 PM com.sap.fooPackage.Node.announce [main] Path: Exiting method

The following 4 sections will further explain each step in details. Bear in mind that in this section
4.2, we try to focus only on the most basic scenario and its API and avoid jumping into the
advanced logic. This should be good enough to get you started. But you will find frequent
references to later sections that cover the logic or advanced features in details.

4.2.1 Identify the output source: Location or Category

Both Location and Category are subclasses of LogController.
Location is the source area that generates trace messages. Typically, it corresponds to the
source code structure, and can be attached to the level of component, package, class or
method.
Category is the source area that generates log messages, corresponding to a distinguished
problem area, such as networking, database,
Naming of the source
Although the naming of a Location and Category is quite flexible, the normal rule is: a valid
hierarchical naming convention.
E.g. Location: com.sap.fooPackage.Class1
com.sap.fooPackage.Class2 (make sure there is no typo)
Category: /System/Networking
/System/Database
(always starts with /)
A common naming practice for Location is, name the location exactly with the full path of Java
package name. Certainly you can randomly name a location with testLocation1_class1 for the
class com.sap.fooPackage.Class1, but the output will not be very meaningful.
Various access API
A static method is provided for each class for easy access to a location or category:
Location.getLocation(<name of the Location>);
Category.getCategory(<name of the Category>);
Tutorial Logging & Tracing Mechanism in SAP
By: Infrastructure Management Team

Or instead of passing the name manually, for Location, you may as well pass the class instance
(java.lang.Object) or the class itself (java.lang.Class). Refer to the Javadocs for details.
In this case, the location object is by default referring to the class level, while using the string
argument (java.lang.String), you have the flexibility in the definition, e.g. include also the
method name to explicitly control logging over methods individually.
Now that you have got a handle to the source, you are ready to configure and ask this source to
generate messages. It is recommended that initially, you should assign make the handle to be
static to improve the efficiency:
static final Location loc = Location.getLocation(this.getClass())

4.2.2 Assign severity to source

Again, refer to Appendix A for various severity level definitions. Simply make use of the
constants provided directly: Severity.DEBUG, Severity.PATH, Severity.INFO, Severity.WARNING,
Severity.ERROR, Severity.FATAL, Severity.ALL and Severity.NONE.
Normally, you assign severity to the source by:
loc.setEffectiveSeverity(Severity.INFO);

Then, any messages with severity lower than INFO will be discarded, others will be directed to the
destination.
Introduce concept of hierarchical severity
Hierarchical severity will be mentioned in Section 4.4.2, but one quick note is, because of the
hierarchical naming feature of location or category, you can save some effort by assigning severity
to the parent, and its children will automatically inherit the assigned severity as well.
static final Location loc = Location.getLocation(com.sap.fooPackage);
loc.setEffectiveSeverity(Severity.INFO);

If you have a location com.sap.fooPackage.Class1, it will have inherited the severity INFO
already.
Default with strictest severity
By default, source object (assume the ascendants has not been assigned severity yet) has
SEVERITY.NONE. Therefore, developer can freely enable the output methods in the source code,
but the actual logging is not activated until it is explicitly switched on when things are ready.

4.2.3 Specify output destination

We need output destination to print out the messages. It is collectively called Log in our tool.
Intuitively, you assign a Log to a Location or Category. Otherwise, without a log, even though you
would have set the severity and inserted the output methods correctly, nothing will be printed. You
can assign a log to a source by using:
loc.addLog(<log1>);

Types of log
Currently, there are 2 types of logs: ConsoleLog, FileLog.
ConsoleLog is for printing to terminal, while FileLog is for writing messages to files.
Multiple destinations
Tutorial Logging & Tracing Mechanism in SAP
By: Infrastructure Management Team

There are times you may want to assign multiple logs to a single source. Message generated from a
source will be sent to both of them simultaneously. For example, output messages to both the
console and a file.
The simplest ways to do so is as below:
loc.addLog(new ConsoleLog());
loc.addLog(new FileLog(C:\\temp\\testOutput.log);

// filepath has to be valid

Creating a log
As you can see from the Javadoc API, the constructors are overloaded where you can specify
certain options for the logs, while we only show the most basic one here (using minimal
parameters for simplicity). Therefore, default settings are used and worth noted:
ConsoleLog is using a TraceFormatter by default.
FileLog is using ListFormatter and the output file will not be a rolling file type, but one single
output file, increasing in size when more and more messages appended to it. More
configurations can be specified for a FileLog and the details are described in 4.4.5.
For example, if you want to switch to a XMLFormatter of a Filelog, you may call:
loc.addLog(new FileLog(C:\temp\testOutput.log, new XMLFormatter());

or with an existing log:

<filelog>.setFormatter(new XMLFormatter());

Attaching a log
Refer to the Javadoc, there are two other APIs in assigning Logs to a Location or Category,
addPrivateLog, addLocalLog, because there are actually three forms of log assignments.
They behave differently in terms of inheritance by children of a Location or Category source. In
short, these three are mutually exclusive, and the example addLog shown here is
straightforward for beginners. It supports forced inheritance to children and message output
unconditionally. More details can be found in sections 4.4.3 and 4.5.1.

4.2.4 Enable output messages

Now that we have the source defined, severity assigned and destination specified properly, you are
ready to insert the output statements in the appropriate places in the code.
Abundant output method APIs
Do not get overwhelmed by the seemingly endless list of output methods. They are not randomly
overloaded, but are expanded from the few fundamental ones as described in Section 3.2. You will
soon realize the pattern in overloading the signatures is fairly systematic. They are simply heavily
overloaded to provide flexibility for the users.
Plus, the set of output methods are very similar between Location and Category.
At this stage, we will not introduce all the overloaded methods, but focus on a few basic and
practical one. (The rest involves better understanding of the logging features, and these are
explained in Section 4.5.2).
Lets divide the output methods into 3 main groups for the ease of explanation:
Typical message output with severity
Denoting the flow of program
Master gate

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

4.2.4.1

Message output with severity

Location

fatalT, errorT, warningT, infoT, pathT, debugT

Category

fatalT, errorT, warningT, infoT

The names are self-explanatory about the severity level of the messages generated by these
methods. The overloaded pattern is the same for each severity output (masked with xxxx
here):
Location
Category
xxxxT(String message)
xxxxT(String subloc, String message)
xxxxT(String message, Object[] args)
xxxxT(String subloc, String message,
Object[] args)

xxxxT(Location loc, String message)

xxxxT(Location loc, String subloc, String message)
xxxxT(Location loc,
String message, Object[] args)
xxxxT(Location loc,
String subloc, String message, Object[] args)

There exists a pattern in method overloading: evolves around the core argument: message.
The addition of subloc, args offers the flexibility for developers to log messages in the
level of details that they need. Understanding these arguments can help you select the
heavily overloaded methods easier.
loc:
It is obvious that the only difference in the API between Location and Category is an
additional loc argument in Category output methods. It is a typical request that log
messages are always written with respect to a source code area. This proves to be very
helpful for logging analysis. By specifying the loc argument, you indicate that the message
should be written as a trace message associated with the loc object. With a little
configuration done for loc, logging can be just done once but will be piped for both
message types (logs & traces) simultaneously. This is explained in details in Section 4.5.2.
Vice versa, this works for Location as well, and API is available to specify the category
argument. But since this is optional for Location, so we wont list them here. (Again, refer
to Section 4.5.2).
subloc:
Treat the subloc argument as the method name of the source class that the message is
generated from. This is optional, but with this argument included in the trace/log, the
picture when doing analysis will be much clearer, especially you can specify different
arguments for overloaded methods.
message:
The actual message to be printed is put in argument message. Make up something simple
that is meaningfully describing the situation/problem. (Aforementioned, language
independency is supported, using ResourceBundle. Refer to 4.4.7 for more examples of
output methods).
args:

Array of additional arguments that are informative, e.g. dynamic information to be included
in the message. This is achieved by using java.text.MessageFormat API to resolve
arguments.

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

To reiterate, refer to the sample code at the beginning of Section 4.2.3, with method
announce(Object o). This is a simple example of working on Location object.
package com.sap.fooPackage;
import com.sap.tc.logging.*;
public class Node {
private static final Location loc =
Location.getLocation("com.sap.fooPackage.Node");
public void announce(Object o) {
String method = "announce(java.lang.Object)";
try {
// do something...eg. connecting to DB, perform certain actions
loc.debugT(method, "Connecting to .);
//minor error in writing something
loc.warningT(method,
"Problems in row {0} to {1},
new Object[] {row1, rowN});
//finish successfully
loc.infoT(method, "DB action done successfully);
}
catch (Exception e) {
}
} // method announce
} // class Node

Potential output, assuming the simplest case with ConsoleLog and default TraceFormatter:
May 3, 2001 6:54:18 PM com.sap.fooPackage.Node.announce [main] Debug: Connecting to .
May 3, 2001 6:54:18 PM com.sap.fooPackage.Node.announce [main] Warning: Problems in row 15 to
18
May 3, 2001 6:54:18 PM com.sap.fooPackage.Node.announce [main] Info: DB action done successfully

Another example of working on Category object:

package com.sap.fooPackage;
import com.sap.tc.logging.*;
public class Node {
private static final Location loc =
Location.getLocation("com.sap.fooPackage.Node");
private static final Category cat =
Category.getCategory("/System/Database");
public void store() {
try {
// Write object data to database ...
}
catch (FailedRegistrationException e) {
cat.errorT(loc,
"store()",
"Error storing node {0} in database.",
new Object[] {this});
}

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

} // method store
// class Node

Note that the output will be identical to the previous example, assuming the default setting
is used (with ConsoleLog and default TraceFormatter). Only when you configure the
output display accordingly (e.g. use another formatter, or change the pattern of the
TraceFormatter), then you will get different result. See Section 4.4.1 for more information.
4.2.4.2

Program flow

This is only available in Location. Tracing the flow of program is a common practice:
entering, exiting, throwing, assertion.

entering: output a default message (Entering Method with Path severity)

indicating that it is entering a source block. NOTE: ALWAYS paired up with
method exiting.
Method
Description
entering()
Entering a source block in general
entering(String subloc)
Specify the method name in subloc
entering(Object[] args)
A general source block with arguments:
Entering method with <args>
entering(String subloc, Object[] args)
Same as above but with specific method name

exiting: output a default message (Exiting Method with Path severity) indicating
that it is leaving a source block. NOTE: ALWAYS paired up with method
entering.
Method
Description
exiting()
Exiting a source block in general. As long as
the methodname (subloc) is specified in
entering , it is not necessary to provide
subloc as argument here anymore. See the
result of the following sample code.
exiting(String subloc) //DEPRECATED
Specify the method name in subloc
exiting(Object res)
A general source block with result:
Exiting method with <res>
exiting(String subloc, Object res)
Same as above but with specific method name

To reiterate, refer to the sample code with method announce(Object o):

public void announce(Object o) {
String method = "announce(java.lang.Object)";
loc.entering(method);
try {
}
catch (Exception e) {
}
loc.exiting();
}

Always log entering

& exiting methods
together.

Potential output, assuming the simplest case with ConsoleLog and default TraceFormatter:
May 3, 2001 6:54:18 PM com.sap.fooPackage.Node.announce [main] Path: Entering method
May 3, 2001 6:54:18 PM com.sap.fooPackage.Node.announce [main] Path: Exiting method

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

throwing: warning message (Throwing ) indicating that the source block is

about to throw an exception
Method
Description
throwing(Throwable exc)
About to throw the exception exc
Throwing(String subloc, Throwable exc)
Same as above but with specific method name

assertion: to test a condition and output an error message, normally with the
assertion included (Assertion failed: <assertion test>) when the evaluation is
false
Method
Description
assertion(Boolean assertion, String desc)
Evaluate the assertion, if false, print desc with the
default message: Assertion failed: <desc> where
<desc> is the assertion test itself, e.g. 5 > 3
assertion(String subloc,
Same as above but with specific method name
Boolean assertion, String desc)

To reiterate, refer to the sample code with method announce(Object o):

public void announce(Object o) {
String method = "announce(java.lang.Object)";
loc.assertion(method, 5<3, "Stupid comparison);
try {
}
catch (Exception e) {
loc.throwing(method, e);
}
}

Potential output, assuming the simplest case with ConsoleLog and default TraceFormatter:
May 3, 2001 6:54:18 PM com.sap.fooPackage.Node.announce [main]
Error: Assertion failed: Stupid comparison
May 3, 2001 6:54:18 PM com.sap.fooPackage.Node.announce [main]
Warning: Throwing java.io.FileNotFoundException:
C:\Not_Exist\zzzzz.log (The system cannot find the path specified)

4.2.4.3

Master gate

LogT

All the other output methods of first type (with severity) are ultimately routed through
this method to actually perform any logging.
Location: basically, these are the same as the first type of method, xxxxxT( ), but
only with an additional severity argument at the beginning:
logT(int
logT(int
logT(int
logT(int

severity,
severity,
severity,
severity,

String
String
String
String

message)
subloc, String message)
message, Object[] args)
subloc, String message, Object[] args)

Category: (same situation as Location):

logT(int severity, Location
logT(int severity, Location
logT(int severity, Location
logT(int severity, Location
Object[] args)

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

loc,
loc,
loc,
loc,

String
String
String
String

message)
subloc, String message)
message, Object[] args)
subloc, String message,

4.3

Typical Practice
The steps and logic of doing logging should be fairly clear by now.
Certainly, the information given is adequate to get users started on instrumenting logging in their
code. However, a good style of doing logging is recommended:
Enabling logging in two major steps: initialization and the individual source classes.
Two stages : higher level vs individual level
Applications are normally made up of different components and are deployed in a distributed
environment. Typically, there should exist a common initialization routine for a component.
Logging configuration, such as setting severity, assigning destination logs, can be done at this
stage as well. Then, the actual insertion of trace/log messages will be done in the respective Java
class level at the reasonable points.
A top-down approach is used, that is, defining coarse configuration at a higher level (componentoriented parent node). Most of the time, users do not need to fine-tune the logging behavior in
such details for each class. Then, a basic configuration at the top level will be good enough to get
logging started. Only when specific setting has to be done individually for certain classes, then the
users can as well enable that in the class level.
Component level at initialization
For example, a component ComX under SAP may have the following package hierarchy:
com.sap.comX.xxx.xxxx.xxx. During initialization, if logging is desired for monitoring important
messages (with severity error or above) to be shown at the console, one should set up something
like this:
Location _loc = Location.getLocation(com.sap.comX);
_loc.setEffectiveSeverity(Severity.ERROR);
_loc.addLog(new ConsoleLog());

Class level
In the individual class, it is recommended to have a static variable to hold the access to the
location object (for better performance) and then start inserting messages:
static Location _loc = Location.getLocation(<classname>.class);

_loc.infoT(.);
_loc.errorT(.);

The code will be cleaner with this 2-level setting. And the reason why this works is because of the
hierarchical feature implemented in the logging framework. Refer to Section 4.4.2 and 4.4.3 for
more details.
You will gradually find that the actual logging part within each class is pretty static, only the
configuration part to manipulate the logging behavior is dynamic, such as, output volume, output
destination, output format. This will be covered in details in Section 4.6.
These two sections provide the basic concepts and show you the basic steps to get you started. In
order to benefit the most out of the tool, you should explore around with the references of the next
few sections.

4.4

Basic Features
This section describes a few tricks that enhance the intelligence to meet the common logging
requirements.

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

4.4.1 More configuration options

Refer to Figure 2-1, it shows the options to alter the logging output through the use of formatter,
log and filter. These are the fundamental means to control how the message should be presented,
where it should be written to or whether it should be screened out.
4.4.1.1

Log (destination)

A log represents the destination where the messages to be written to. It is also the object we
will specify and assign an appropriate formatter for the log, as well as multiple optional
filters (see the next two sections). Like the log controller source object, a severity level can
be defined for each log, but unlike the default for log controller (Severity.NONE), the
default for a log is Severity.ALL, that is, no effect on the main severity level check with log
controller. (See Appendix C for an overview of default values.)
Currently, the following logs are provided:
ConsoleLog
Direct the messages to System.err (java.lang.System)
Typical use in debugging process for a quick view of problems
FileLog
Direct the messages to a file or a set of rotating files (more details in Section 4.4.5)
StreamLog
Direct the messages into an arbitrary OutputStream (java.io.OutputStream)
4.4.1.2

Formatter

Each type of log destination can print the messages in different formats. Currently, there
are three major types of formatter:
TraceFormatter
Human readable format
This is very likely to be the most commonly used formatter when users want to
quickly understand what is going on with the application.
Normally, not used with log viewers (refer to Section 6.2). Therefore, users can
customize the pattern of the TraceFormatter, indicated by various placeholders.
Details can be referred to the Javadoc of method setPattern. By default, the
pattern is:
%24d %-40l [%t] %s: %m

and the corresponding output is:

Jan 01, 2001 10:10:00 PM
message

com.sap.FooClass.fooMethod [main] Fatal: A sample fatal

For example, if the pattern becomes:

%24d %l [%s] %m

The corresponding output is:

Jan 01, 2001 10:10:00 PM com.sap.FooClass.fooMethod [Fatal] A sample fatal message

The number in the pattern denotes the width of a field. In the first example: %-40l
indicates a 40 char limit for the location name (com.sap.FooClass.fooMethod) that will
be right aligned (with a minus sign). If the length of a string exceeds the width
defined, the string will be truncated. The second example: %l simply displays the
full string that will be left aligned.
The meaning of the place holders (details can be found in Javadoc):
Placeholder Description
%d
timestamp in readable form

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

%l
%c
%t
%s
%m
%I
%p
%g

the location of origin

(eg: com.sap.FooClass.fooMethod)
the log controller the message was issued through
(eg: com.sap.FooClass)
the thread that emitted the message
the message severity
the formatted message text
the message id
the time stamp in milliseconds since January 1, 1970
00:00:00 GMT
the group identification

Table 4-1 Placeholders of TraceFormatter

XMLFormatter
Suitable for file transfer to be further processed in other applications
Sample out of the same result is:
<record>
<id>10.48.27.165:4A5AB2:E99D2EDAFF:-8000</id>
<time>Mon Jan 01 22:00:00 PDT 2001</time>
<source>com.sap.FooClass</source>
<location>com.sap.FooClass.fooMethod</location>
<thread>main</thread>
<severity>Fatal</severity>
<msg-type>Plain</msg-type>
<msg-clear>A sample fatal message</msg-clear>
</record>

The DTD of the result is:

<!ELEMENT table (id, time, source, location, thread, group?, severity, relatives?, msgtype, msg-code?, bundle?, msg-clear?, args?)>
<!ELEMENT id (#PCDATA)>
<!ELEMENT time (#PCDATA)>
<!ELEMENT source (#PCDATA)>
<!ELEMENT location (#PCDATA)>
<!ELEMENT thread (#PCDATA)>
<!ELEMENT group (id, level, indent)>
<!ELEMENT level (#PCDATA)>
<!ELEMENT indent (#PCDATA)>
<!ELEMENT severity (#PCDATA)>
<!ELEMENT relatives (relative+)>
<!ELEMENT relative (#PCDATA)>
<!ELEMENT msg-type (#PCDATA)>
<!ELEMENT msg-code (#PCDATA)>
<!ELEMENT bundle (#PCDATA)>
<!ELEMENT msg-clear (#PCDATA)>
<!ELEMENT args (arg+)>
<!ELEMENT arg (#PCDATA)>

ListFormatter
Output in this format serves like a router to send the data to be processed further by
another application, e.g. a log viewer, instead of being read directly by an end-user
As simple hash-separated fields to be processed mainly with a log viewer
Sample output of the same result is, as of version 1.3:
#1.3#10.48.27.165:4A5AB2:E99D42D4F4:-8000#Mon Jan 01 22:00:00 PDT
2001#com.sap.FooClass#com.sap.FooClass.fooMethod#main##0#0#Fatal##Plain###A
sample fatal message#

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

 The order of the fields, delimited by the hash sign is:

o Version (of the ListFormatter)
o Message id
o Timestamp
o Source name (log controller)
o Location name (the actual code location that generates the message)
o Thread id
o (Group id)
o Group level
o Group indentation
o Message severity
o (Relatives names: can be multiple)
o Message Type (PLAIN or JAVA)
o (Message Code)
o (ResourceBundle name)
o Message
o (Number of arguments)
o (Arguments)
4.4.1.3

Filter

Refer to Figure 2-1, filter can be added to both LogController(source) and/or

Log(destination) to further restrict or alter the output behavior.
Multiple filters are allowed to assign to each LogController and/or each Log.
Implementation by users
Actual implementation of filter has to be done by users, according to the interface defined
in the package. Basically, this evaluates the LogRecord object and veto accordingly. The
name of the only method indicates that, a boolean true is returned when the logRecord
passes the filter, otherwise false.
Since the implementation is done by users, classname of the filter will not be known in
advance. If the user-defined filter class is to be specified in the configuration file, a fully
qualified classname should be used. Be careful that this only works if the class is already
included in the classpath during the configuration. Should it needs its own classloader, the
latter has to be set clearly in the configurator constructor. Details of configuration tool can
be found in Section 4.6.

4.4.2 Hierarchical severity inheritance

By now, you should be familiar with the severity scale defined in this tool. In Section 4.2.2, you
learn how to assign severity to a source (location or category). Normally, there are numerous
source objects in your program, and the task will become very tedious if severity has to be
assigned individually to each of them. Thus, the requirement of a hierarchical naming convention
(as mentioned in Section 4.2.1) has its reason now.
Interfere inheritance
Intuitively, severity assigned to the parent node will be effective for its children and all
descendants. By and large, this is true. But there is little more that allows you to control the
inheritance behavior. Descendants do not need to inherit the severity from parent with the
following three ways:

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

 Explicit severity setting (default: Severity.NONE)

As long as the child object has been explicitly assigned a severity, this will win, and will not be
affected by the severity assignment of its ascendants. This corresponds to the API:
static final Location loc = Location.getLocation(com.sap.foo);
loc.setEffectiveSeverity(Severity.INFO);

So, if parent com.sap is assigned Severity.FATAL, this has no influence on com.sap.foo,

the latter still has Severity.INFO assigned.
By the same token, you can easily deduce that descendants (without explicit severity assignment)
will inherit the severity from the closest ascendants that has explicit effective severity assigned.
With the same example shown above, what will be the severity of com.sap.foo.child1?
FATAL or INFO?
Remember, all descendants of com.sap.foo will now inherit severity from com.sap.foo.

Define a severity floor limit: Minimum severity (default: Severity.ALL)

Assigned severity should not be more relaxed than this
Define a severity ceiling limit: Maximum severity (default: Severity.NONE)
Assigned severity should not be stricter than this
You can specify the floor and ceiling thresholds individually or together. These two define the
valid severity range for a source object. Therefore, by default, it does not restrict any inherited
severity because it must fall within the default minimum severity ALL and maximum severity
NONE.
Example
Consider the following, where explicit severity has NOT been set for com.sap.foo, but the
range has been defined as:
static final Location loc = Location.getLocation(com.sap.foo);
loc.setMinimumSeverity(Severity.DEBUG);
loc.setMaximumSeverity(Severity.WARNING);

When parent com.sap is explicitly assigned with Severity.FATAL, this severity will not be
inherited straightly by com.sap.foo because it gets denied by its ceiling threshold. Instead,
com.sapmarkets.foo will get the ceiling threshold Severity.WARNING. The same theory
applies to the floor threshold, i.e., any severity of parent that is lower than Severity.DEBUG will
be ignored.
Only when the parents severity falls between the severity range of the child can overwrite the
childs severity with its own value.
(A technical side note: assigning explicit severity, setEffectiveSeverity(Severity.INFO),
automatically overwrites the value of both MinimumSeverity and Maximum Severity into
Severity.INFO).
Lets try to organize the possibilities here:
A default setting:

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Location/
Category

Maximum
Severity

com.sapmarkets
com.sap.foo1
com.sap.foo1.child1
com.sap.foo2
com.sap.foo2.child1

NONE
NONE
NONE
NONE
NONE

Minimum
Severity
ALL
ALL
ALL
ALL
ALL

Resulting
(Effective)
Severity

Description

NONE
NONE
NONE
NONE
NONE

default
default
default
default
default

Scenario 1:
Location/
Category

Maximum
Severity

Minimum
Severity

Resulting
(Effective)
Severity

Description

com.sap

WARNING

WARNING

WARNING

com.sap.foo1

FATAL

ALL

WARNING

setEffectiveSeverity =
WARNING
setMaximumSeverity =
FATAL

com.sap.foo1.child1
com.sap.foo2
com.sap.foo2.child1

NONE
NONE
INFO

ALL
ALL
ALL

WARNING
WARNING
INFO

setMaximumSeverity =
INFO

Scenario 2:
Location/
Category

Maximum
Severity

Minimum
Severity

Resulting
(Effective)
Severity

Description

com.sap

INFO

INFO

INFO

com.sap.foo1

FATAL

FATAL

setEffectiveSeverity =
WARNING
setEffetiveSeverity =
FATAL

com.sap.foo1.child1
com.sap.foo2
com.sap.foo2.child1

NONE
NONE
FATAL

ALL
ALL
WARNING

FATAL
FATAL
INFO
WARNING

setMinimumSeverity =
WARNING

4.4.3 Hierarchical destination inheritance

The naming hierarchy does not only enable inheritance severity, but also destination (log)
assignment. The logic is even more straightforward here:
static final Location parent = Location.getLocation(com.sap),
Location child =
Location.getLocation(com.sap.foo);
parent.addLog(new ConsoleLog());
parent.fatalT(A fatal message from parent);
child.fatalT(A fatal message from children);

Both messages will be output to the console, even ConsoleLog is assigned only to the parent node.
Additivity

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Multiple destination logs are allowed. So, unlike the severity inheritance logic, the assignment of
logs to a source object is additive. There is no race condition. For example, adding to the previous
code (original code greyed out):
static final Location parent = Location.getLocation(com.sap),
Location child =
Location.getLocation(com.sap.foo);
parent.addLog(new ConsoleLog());
parent.fatalT(A fatal message from parent);
child.fatalT(A fatal message from children);
child.addLog(new FileLog(<a file>));

In the Console, there will still be two messages printed, each from parent and child, while in the
output file <a file>, messages from child (here, only 1 message) are also printed, in addition to
Console destination.
One might want to suppress this inheritance feature in certain circumstances to fine-tune the output
behavior. This is doable, as you can find more details in Section 4.5.1.

4.4.4 Severity evaluation

Whether a message can be logged or not, largely depends on its severity level compared against
the severity threshold assigned to the source object (another factor is the use of filter, if any). In
some cases, developers do not feel comfortable in simply calling the output API (logT, fatalT, errorT,
) due to potential performance issue, for example, resolving the parameter list of the call. An
additional severity check before making the output calls is desired.
There are APIs allow you to do this. You can find them under the class LogController. They
correspond to the output APIs and they are:
beLogged(Severity level)
beFatal(), beError(), beWarning(), beInfo(), bePath(), beDebug()

These return a boolean true if the message passes the respective severity level check, otherwise
false.

4.4.5 Output File

Directing messages into a file is a very popular practice. In this tool, this is done by assigning
FileLog to your source objects. In fact, the Javadocs has very clear and detailed instruction on this.
This section will recap and explain certain common configuration features to complement the
Javadocs.
Refer to Appendix C for the summary of default behavior. We have mentioned the FileLog class
twice in the two tables. By default, messages written to an output file will be in the ListFormatter
format, without using any special character encoding. And there will only be one single output file,
increasing in size constantly. There are number of options that you can configure the behavior of
an output file, via API or configuration file (refer to Section 4.6).
Filename
Limit file size and do sequencing on the output file
FileLog vs Physical file

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

4.4.5.1

Filename

Filename can be expressed as the fully qualified filepath, such as C:\temp\trace.log, or

expressed in pattern with a number of available placeholders. This can take care of the
potential hardcoding problem and platform dependent issue, such as filename separator.
Place holder

Description

Comments

Local file name separator

%h
%t

Home directory
System temporary directory

%u

Unique number to make the file

name unique
The percentage sign
Sequence number of the file

C:\temp\trace.log == C:/temp/trace.log but

probably you would like to use the latter
Value of system property: user.home
Value of system property: java.io.tmpdir.
Normally, it is C:\temp. Therefore:
C:\temp\trace.log == %t/trace.log
GUID. To make sure no naming conflict will occur.

%%
%g

This is useful only when users specify the file size

limit and the max. number count to do rotation. By
default, this will be appended to the end of the
filename/filepattern specified. E.g.
%t/trace.%g.log ----> C:\temp\trace.0.log
(otherwise, C:\temp\trace.log.0 by default)

Table 4-2 Placeholders of filename

4.4.5.2

Output file sequence and rotation

This is implicitly done when users have specified values for the filesize (in byte) and the
cap for the sequencing count before rotating back to the first file. This pair of parameters
has to coexist together. Either they both have values assigned or both values equal to zero.
Otherwise, depends the option you are using to do this configuration, either an exception
can be thrown (through programming API) or values with best guess will be used (through
configuration file, see Section 4.6), e.g. both attributes get default zero values (refer to
Section 5.1 for this fault-tolerance logic).
This can be done when calling the FileLog constructor:
FileLog

_file = new FileLog(%t/trace.%g.log,

800000,
10,
new TraceFormatter());

Or a quick peek on the configuration file syntax (refer to Section 4.6):

log[File]
log[File].pattern
log[File].limit
log[File].cnt
log[File].formatter

=
=
=
=
=

FileLog
%t/trace.%g.log
800000
10
TraceFormatter

The first output file will be created C:\temp\trace.0.log (NOTE: sequence number starts
with 0), and when its size approaches and then exceeds the limit (800,000 bytes), the next
incoming message will be directed to a new file C:\temp\trace.1.log. This process keeps
going until the last sequenced file C:\temp\trace.9.log (count = 10), then the next file to
be written will again become C:\temp\trace.0.log.
4.4.5.3

FileLog vs physical file

A FileLog is a logical representation for a physical output file where messages are directed
to. With the several configuration options mentioned in the previous section, users can
Tutorial Logging & Tracing Mechanism in SAP
By: Infrastructure Management Team

manipulate the output behavior of the output file through the API of FileLog, as well as do
(multiple) assignment(s) of the FileLog to (various) source object(s).
1:1 recommended
Intuitively and normally, this should be a 1:1 relationship between the FileLog and the
actual output file. It is rare that a user will want to create two instances of FileLog on the
same file and with different configuration, e.g. one with TraceFormatter and the other one
with XMLFormatter. Currently, this is not illegal. But it does not look like a reasonable
design, and user will run into the risk of improper synchronization of messages in a multithreading environment. The full message text may get interrupted and interweaved with
another message text.
Therefore, this is NOT recommended. At the meantime, please pay attention to this when
coding or doing configuration.

4.4.6 Change Output Logs/Filters

The decoupled design of the logging framework makes it flexible to configure the logging
behavior. One example is to change the log destination assignment to a source object. As
mentioned in Section 4.4.3, the assignment of logs to the source objects is additive. This applies to
filter assignment too.
That is, adding an additional output destination is straightforward, but if you need to switch the
output destination, e.g. from a console to a file, user must include instructions to remove the
ConsoleLog from the source object and then assign a FileLog to it. The APIs that support this:
then,

_loc.removeLog(<a specific log>);

_loc.addLog(<the new log>);

For filters, the APIs are similar:

then,

_loc.removeFilter(<a specific filter>);

_loc.addFilter(<the new filter>);

Eventually, update of this kind is very likely to be done with configuration file (Section 4.6), and
changes to logging properties during runtime are supported. There is a special syntax to deal with
additive attributes. Refer to the end of section 4.6.2.3. A quick peek again:
log[File]
log[File].pattern
log[File].formatter
com.sap.logs

= FileLog
= %t/trace.log
= ListFormatter
= log[File]

## (1) in the next round, user could switch the output to Console:
com.sap.logs
= ConsoleLog
## (2) or you can add additional console ouput:
com.sap.logs
= + ConsoleLog

4.4.7 Language Dependency

Mainly for log messages
This is rarely an issue for trace messages that mainly used by developers during the debugging
process. On the other hand, log messages, which are related to specific logical areas, are most
likely to be read by administrators, end-user support group, consulting group to diagnose
application problem during runtime. Not only a meaningful text should be logged, but also a
message translation mechanism should be supported as application may be deployed worldwide.
Tutorial Logging & Tracing Mechanism in SAP
By: Infrastructure Management Team

ResourceBundle
This logging tool does support language dependency with the use of ResourceBundle provided in
Java. (For more background about this, please refer to the java class java.util.ResourceBundle).
The basic concept is, the message text logged by the developers in their program is neutral. It is
like a message code/id and a resource bundle file for each language exists to map the message
code into the text of the respective language. During the application runtime, depends on the
locale, resource bundle file will be looked up, and message text will be translated and localized
accordingly.
More APIs
To meet this requirement, there exists a few more output methods API under Category class to
enable the translation feature for the log messages. These do not appear under the Location class
(in terms of the basic ones you have learnt in Section 4.2.4.1 that do not involve the reference to
category).
Recall the output methods of Category you have learnt in Section 4.2.4.1 and 4.2.4.3. If you are
familiar with them, you should have no problem in using these additional output methods to enable
translation, because they are very similar and have the same overloading pattern. Basically, these
are methods
having identical corresponding method names, but ended without a T
replacing the argument String message into Object msgCode
The table below shows a few commonly used methods, compared to the classic ones that you
have learnt previously.
Category (xxxx stands for the severity level)

Translation supported

xxxxT(Location loc, String message)

xxxxT(Location loc, String subloc,
String message)
xxxxT(Location loc,
String message, Object[] args)
xxxxT(Location loc, String subloc,
String message, Object[] args)
logT(int severity, Location loc, String message)
logT(int severity, Location loc, String subloc,
String message)
logT(int severity, Location loc, String message,
Object[] args)
logT(int severity, Location loc, String subloc,
String message, Object[] args)

xxxx(Location loc, Object msgCode)

xxxx(Location loc, String subloc,
Object msgCode)
xxxx(Location loc,
Object msgCode, Object[] args)
xxxx(Location loc, String subloc,
Object msgCode, Object[] args)
log(int severity, Location loc, Object msgCode)
log(int severity, Location loc, String subloc,
Object msgCode)
log(int severity, Location loc, Object msgCode,
Object[] args)
log(int severity, Location loc, String subloc,
Object msgCode, Object[] args)

Table 4-3 Common Output API with support of translation

Sample
MsgCode is the message id specified in the resource bundle.
For example, there are two resource files; we have the following translation (just to make fun of
British English and American English). MsgCode is the key specified in the left of the pair.
(1)
DBProblem
ProcessCanceled
WrongColor

= Problem with Database.

= Process {0} canceled.
= Wrong Color: {0}.

DBProblem

= Problem with Database.

(2)

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

ProcessCanceled
WrongColor

= Process {0} cancelled.

= Wrong Colour: {0}.

In the program, the code becomes:

_cat.error(_loc, ProcessCanceled);

During runtime, depends on the specified locale (or the default locale of the platform), the output
will be either:
Process #### canceled.
Or
Process #### cancelled.
Invalid translation
What if the MsgCode specified is invalid? A MissingResourceException will be thrown. In
this case, the exception will be caught and logged by the logging framework itself without
jeopardizing the flow of the application. The value string of MsgCode itself will be used as the
translated string. This is the best substitute in this situation. (This is actually the fault-tolerance
mechanism provided by the logging tool. See Section 5.1 for details).
There exists another option that developer can specify a meaningful message string as a backup to
get around with the invalid MsgCode problem. Then, the logging framework needs not guess a
substitute value, and this backup string can give more meaningful message (although nontranslated) than a MsgCode. This is supported by the use of an additional argument MsgClear.
Note that the same set of method APIs is overloaded again with this additional argument (method
overloading is actually very systematic, so, dont get daunted by the number of methods).
Fallback supported
xxxx(Location loc, Object msgCode, String msgClear)
xxxx(Location loc, String subloc, Object msgCode, String msgClear)
xxxx(Location loc, Object msgCode, Object[] args, String msgClear)
xxxx(Location loc, String subloc, Object msgCode, Object[] args, String msgClear)
Log(int severity, Location loc, Object msgCode, String msgClear)
Log(int severity, Location loc, String subloc, Object msgCode, String msgClear)
Log(int severity, Location loc, Object msgCode, Object[] args, String msgClear)
Log(int severity, Location loc, String subloc, Object msgCode, Object[] args, String msgClear)

This is a fallback option when the use of MsgCode is not very reliable value. When argument
MsgClear (non-null) is available, its value always exists in parallel with the MsgCode. This option
becomes particularly useful when log viewers are involved. In most cases, the assumption is
MsgCode will be used by log viewer and translation is done at the time of viewing whenever the
MsgCode is valid, otherwise, MsgClear can be used as a backup.

4.5

Advanced Features

4.5.1 Three forms of log assignment

Normally, users like to take advantage of the inheritance property to simplify the configuration
task, e.g. for both severity and log assignment. The assignment of log is additive, and at times,
more control over the log assignment is needed to fine-tune the specific assignment for certain
source objects at certain levels.
There actually exist three kinds of log which you can assign to a source object:
Tutorial Logging & Tracing Mechanism in SAP
By: Infrastructure Management Team

 common log
local log
private log
They are mutually exclusive for each source object. The assignment of one type will
automatically disable the assignment of the other type (if any).
4.5.1.1

Common Log

So far, this is the one that has been introduced previously. This allows regular inheritance;
this log will become available to all descendants of the parent object.
The corresponding API is:
static final Location parent =
Location.getLocation(com.sap),
Location child =
Location.getLocation(com.sap.foo);
parent.addLog(new ConsoleLog());
parent.setEffectiveSeverity(Severity.NONE);
child.setEffectiveSeverity(Severity.INFO);
child.fatalT(A fatal message from children);

This is an unconditional inheritance. As long as the message passes the evaluation of

severity and filter of the child object, it will be printed out via the inherited log, in this case,
ConsoleLog.
4.5.1.2

Local Log

Inheritance is allowed, but with additional condition: final discretion of message printing
lies with the original parent object. The log is local, in the sense that it is not available for
the descendants if the child message does not pass the severity and filter test of the parent
object.
static final Location parent =
Location.getLocation(com.sap),
Location child =
Location.getLocation(com.sap.foo);
parent.addLocalLog(new ConsoleLog());
parent.setEffectiveSeverity(Severity.NONE);
child.setEffectiveSeverity(Severity.INFO);
child.fatalT(A fatal message from children);

Contrary to the previous example, the child message will not be printed because it does not
pass the severity test of the parent.
4.5.1.3

Private Log

This completely disables inheritance. Log assignment is private: only effective for the
parent source object.
static final Location parent =
Location.getLocation(com.sap),
Location child =
Location.getLocation(com.sap.foo);
parent.addPrivateLog(new ConsoleLog());
Tutorial Logging & Tracing Mechanism in SAP
By: Infrastructure Management Team

child.setEffectiveSeverity(Severity.INFO);
child.fatalT(A fatal message from children);

In this case, child object does not inherit any destination log at all. Nothing will be printed
even the message passes all the severity and filter tests.

4.5.2 Relations between Category & Location

(This section and the next are closely related.)
Reason
It is a common practice to look at the log messages and trace messages together when doing
diagnosis. A correlation between a problematic logical area and the source code location that
generates the problem is highly desired. For example, an error occurs when closing the database,
the actual location of the source code (from which class, which method, with what argument(s)) is
reported as well.
You have already seen the introduction of how this can be supported in Section 4.2.4.1 to generate
both log and trace messages in parallel, with the use of category and location.
We will elaborate an example here.
Example
For simplicity, we only show the example of attaching a location to a category (we will here refer
to the location being associated to the category as relative of the category), but this is also true for
the vice versa case.
Note that the API supports precisely the assignment of a category to a location and also the other
way round, but not among the same type. It is legitimate to associate more than one category to a
location at each logging, but only one location for one category. Refer to the output method APIs
of each class.
Package com.sap.fooPackage;
import com.sap.tc.logging.*;
public class Node {
private static final Location loc =
Location.getLocation("com.sap.fooPackage.Node");
private static final Category objMgmt =
Category.getCategory("/Objects/Management");
public void announce(Object o) {
final String method = "announce(java.lang.Object)";
loc.entering(method, new Object[] {o});
try {
// Register object ...
}
catch (RegistrationException e) {
objMgmt.errorT(loc,
method,
"Error registering object {0}.",
new Object[] {o});
}

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

loc.exiting();
} // method announce
} // class Node

In order to output all the trace and log messages highlighted in the example above, user will have
the following severity setting, for example:
loc.setEffectiveSeverity(Severity.PATH);
objMgmt.setEffectiveSeverity(Severity.ERROR);
conLog = new ConsoleLog();
loc.addLog(conLog);
objMgmt.addLog(conLog);

Output
And for the output line from the category objMgmt, it will output 2 messages simultaneously:
1 log message and 1 trace message. They will have the same message id for cross-referencing
each other. This makes the analysis more comfortable.
(But recall the information in sections 4.3, 4.4.2 and 4.4.3, the configuration could have already
been taken care of in the initialization stage and is effective to this local scope through the
hierarchical inheritance logic).
If the location has stricter severity setting, e.g. default Severity.NONE, all the trace output will be
suppressed, including the one from the category, that is, that output line will NOT produce two
messages simultaneously, but only the log message.
More control
Naturally, the next question may be whether we can do more advanced configuration regarding the
correlated category and location source objects. This is positive.
Consider Category /Objects/Management; we only want to focus on some extreme situations,
that is, messages with severity FATAL. Several source code location (com.sap.xxx.a,
com.sap.xxx.b, .) can result in a fatal condition in this logical area. And for some reasons,
we are particularly interested in one of them, e.g. com.sap.xxx.a and would like to generate
more output messages, including all with severity INFO or above related with this location only,
while maintaining FATAL for the rest.
Having a tight severity control initially and then relaxing it for particular areas in question to
produce more output for analysis is essential. More details will be shown in the next section,
which shows how to achieve this by playing with the relative severities.

4.5.3 Relative severities

In the previous section, we showed that you could tag along a log controller to another log
controller. In this case, the amount of output can be configured with respect to the relative(s) too.
The evaluation of the message output not only depends on the log controllers own severity level,
but also the relative severity of its relative(s), if any.
4.5.3.1

Set up relative severity (add relative)

The API is really similar to the regular one:

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Regular

Location

Category

setEffectiveSeverity
(int severity)
setEffectiveSeverity()

setEffectiveSeverity
(Category relative, int severity)
setEffectiveSeverity
(Category relative)
setMaximumSeverity
(Category relative, int severity)
setMaximumSeverity
(Category relative)
setMinimumSeverity
(Category relative, int severity)
setMinimumSeverity
(Category relative)

setEffectiveSeverity
(Location relative, int severity)
setEffectiveSeverity
(Location relative)
setMaximumSeverity
(Location relative, int severity)
setMaximumSeverity
(Location relative)
setMinimumSeverity
(Location relative, int severity)
setMinimumSeverity
(Location relative)

setMaximumSeverity
(int severity)
setMaximumSeverity()
setMinimumSeverity
(int severity)
setMinimumSeverity()

Comment
Reset severity (with
respect to the relative)

Reset

Reset

Table 4-4 Setting relative severity

Normally, you still assign severity in the regular fashion, as those listed in column 1. The
typical requirement is to assign a stricter severity in the primary source object, and a more
relaxed level tagged with a certain relative (not the other way round!).
The relative severity is specific to the pair of the location and category. When you want to
change the relative severity or reset it, the relative has to be clearly indicated in the API.
The relative severity is unidirectional:
loc1.setEffectiveSeverity(cat1, Severity.INFO);

The following is not implicitly done:

cat1.setEffecitiveSeverity(loc1, Severity.INFO);

Relative severity is also hereditary. Descendants also inherit the relative severity
assignment. The same logic of local restriction through the use of maximum(ceiling) and
minimum(floor) severity limit (described in Section 4.4.2) is also applicable to the
descendants. Note that evaluation is always done with respective to the relative(s).
4.5.3.2

Evaluation of severity with relatives

The severity of a log controller can be different when a relative severity (or severities) has
been set. The rule of thumb is: inclined to a more relaxed setting.
That is, the minimum severity of the two factors below will always win:
direct severity of itself
severity related to its relative
if there exists multiple relatives, the minimum of the relative severities wins
This applies to all 3 types of severities shown in the table above: effective, maximum, and
minimum.The following tables highlights the severity results in both cases of evaluating
with and without a relative.
Regular:
setEffectiveSeverity
(int severity)

With Relative:
setEffectiveSeverity
(<log controller> relative,
int severity)

Result:
getEffectiveSeverity()

Result:
getEffectiveSeverity
(LogController relative)

FATAL
<not set>
FATAL
INFO

<not set>
INFO
INFO
ERROR

FATAL
NONE
FATAL
INFO

FATAL
INFO
INFO
INFO

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Comments

Not stricter

Regular:
setMaximumSeverity
(int severity)

With Relative:
setMaximumSeverity
(<log controller> relative,
int severity)

Result:
getMaximumSeverity()

Result:
getMaximumSeverity
(LogController
relative)

FATAL
<not set>
FATAL
INFO

<nothing>
INFO
INFO
ERROR

FATAL
NONE
FATAL
INFO

FATAL
INFO
INFO
INFO

Regular:
setMinimumSeverity
(int severity)

With Relative:
setMinimumSeverity
(<log controller> relative,
int severity)

Result:
getMinimumSeverity()

Result:
getMinimumSeverity
(LogController
relative)

FATAL
<not set>
FATAL
INFO

<nothing>
INFO
INFO
ERROR

FATAL
NONE
FATAL
INFO

FATAL
INFO
INFO
INFO

Comments

Not stricter

Comments

Not stricter

As you can see, the use of relative severity is pretty independent. It never affects the
individual severity evaluation of the log controller itself. If users decide not to configure the
output with respective to any relative, the relative severity assignment has no effect on the
output. The result will stay intact as listed in column 3.
4.5.3.3

Output resulting from relative severity

Use the table below to elaborate (effectiveSeverity). What will be the result of the following
code for each of the 4 conditions listed? This is an example to produce log messages with
respect to a location:
Category cat = Category.getCategory(Objects/Management);
Location loc = Location.getLocation(com.sap.xxx.a);
cat.warningT(loc,
methodname,
"Error registering object {0}.");
Regular:
cat.setEffectiveSeverity
(<severity>)

With Relative:
cat.setEffectiveSeverity
(loc, <severity>)

FATAL
<not set>
FATAL
INFO

<not set>
INFO
INFO
ERROR

Result, output enabled

(with relative loc)?
cat.warningT(loc, .)
NO
YES
YES
YES

Another example: to produce traces from location with respect to a category:

Category cat = Category.getCategory(Objects/Management);
Location loc = Location.getLocation(com.sap.xxx.a);
loc.warningT(methodname, messageCode);
loc.warning(cat, methodname, messageCode);
Regular:
loc.setEffectiveSeverity
(<severity>)

With Relative:
loc.setEffectiveSeverity
(cat, <severity>)

FATAL
<not set>

<not set>
INFO

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Result, output enabled

(without relative cat)?
loc.warningT( ..)
NO
NO

Result, output enabled

(with relative cat)?
loc.warning(cat, ...)
NO
YES

FATAL
INFO

INFO
ERROR

NO
YES

YES
YES

When the output methods API denote a relative(s), the logging framework tries to output
the message simultaneously to both the log controller and its relative(s). The evaluations
are independent and done separately. The results shown in the tables above are those for
the log controller itself.
Whether the output is enabled at the relative totally depends on the configuration set by the
relative itself, which at the same time, evaluate the severity with respective back to the log
controller. If no explicit relative severity is configured by the relative, the situation
becomes similar to the conditions listed in the table below. (Log a message from a category
with a relative location).
Regular:
cat.setEffectiveSeverity
(<severity>)

With Relative:
cat.setEffectiveSeverity
(loc, <severity>)

FATAL
<not set>
FATAL
INFO

<not set>
INFO
INFO
ERROR

Output enabled for cat?

cat.warningT(loc, .)
NO
YES
YES
YES

Output enabled by relative

loc?
(no explicit severity setting)
NO
NO
NO
NO

For the case that relative severity is also specified at the relative loc (assuming its own
severity remains the default Severity.NONE value).
Regular:
cat.setEffectiveSeverity
(<severity>)

With Relative:
cat.setEffectiveSeverity
(loc, <severity>)

FATAL
<not set>
FATAL
INFO

<not set>
INFO
INFO
ERROR

4.5.3.4

<not set>
INFO
WARNING
ERROR

Output enabled by relative

loc?
NO
YES
YES
NO

Summary of behavior

4.5.3.5

At loc:
loc.setEffectiveSeverity
(cat, <severity>)

Relative severity assignment is precise with the pair, unidirectional

Not affecting severity of log controller itself
Not affecting the severity of log controller being attached
Therefore, output methods also have to be explicit with which relative(s)
Severity relative is hereditary
Also, explicit relative assignment to be specified for the relative itself, if so desired
Evaluation always results in a more relaxing condition (allows more output)
Evaluation is done independently for the calling log controller and the relative log
controller. Each evaluation treats the other log controller as relative and does the
calculation with respective to its own relative severity configuration (if any).

Alternative for customer-defined severities

This logging framework is providing about 8 severity levels to control the logging output.
In practice, it shows that this number of levels is adequate enough for most applications.
The majority of users mainly use 3 or 4 severity levels. But indeed certain users require
additional severity levels for some special logging behavior, mainly for distinguishing
different applications/stages, e.g., application area X, Y; initialization stage,

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Common requirement
Defining your own severity levels sounds like a straightforward solution, that is, creating a
distinguished severity level to independently control the logging output for a specific
stage/area. For example, during the initialization stage, to output messages that has severity
level between Severity.WARNING and Severity.ERROR. Therefore, a customized severity
level, e.g. SEVERITY_INIT which is stricter than WARNING and lower than ERROR is
desired.
However, there are a few drawbacks with this approach:
Order of severity levels always total
Definition of the customized severity levels is unclear to users, e.g. support group
For developers, it is also difficult to maintain and coordinate the customized
severity levels among components and applications
Solution with SAP API
This issue can be well handled by the SAP logging API which provides support for both
logging (category) and tracing (location). As you have already learnt the details, you
should be aware of the flexibility to adjust severity level independently for different focuses
with the use of relative severity, e.g. more trace messages for initialization stage
initialization. This can be achieved easily by the following line, with category
initialization:
loc.setEffecitiveSeverity(initialization, Severity.WARNING);

Then, when writing message, call with respective to this relative initialization:
loc.errorT(initialization, <.message.>);

Assuming logging is not turned on for others (with default Severity.NONE), trace messages
of severity warning or above, regarding to initialization will still be written to the
corresponding destination logs.

4.5.4 Multi-threading
This logging tool works well in a multi-threading environment.
Thread id is stored with the message, and this can be displayed or masked in the output. Recall the
configuration that can be done for TraceFormatter, where you can specify the pattern of the
format to include the thread id or not.
A sample output in trace format:
Jan 01, 2001 10:10:00 PM
Jan 01, 2001 10:10:00 PM
Jan 01, 2001 10:10:00 PM
Jan 01, 2001 10:10:00 PM
Jan 01, 2001 10:10:00 PM
Jan 01, 2001 10:10:00 PM

4.6

com.sap.fooClass.fooMethod [Thread-0] Fatal: A fatal message

com.sap.fooClass.fooMethod [Thread-1] Fatal: A fatal message
com.sap.fooClass.fooMethod2 [Thread-1] Error: An error message
com.sap.fooClass.fooMethod [Thread-2] Fatal: A fatal message
com.sap.fooClass.fooMethod2 [Thread-0] Error: An error message
com.sap.fooClass.fooMethod2 [Thread-2] Error: An error message

Configuration tool
Configurator

PropertiesConfigurator

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

This is one of the most demanded features.

A means to independently configure the behavior and output of logging, without messing with the
source code is very important. The most common case is, users would want to generate more or
fewer log messages, depends on the diagnosis level they are in. Sometimes, users would also want
to change the output destination or even the format of the message. These all can be specified
separately and dynamically integrated with the program. No recompilation of the source code is
necessary.
If you are clear about the concept, when you refer back to the example shown in Section 4.2, one
can immediately realize that, eventually, what are left in the source code are pretty much the API
of step 1 and step 4. The rest can be taken out and placed in a properties file. The following two
sections will show you how to read the properties file from your code, and how to construct a
properties file.

4.6.1 To use
4.6.1.1

Calling it

Now, on top of using the API in your code to do configuration, like setting severity,
designating output destination, you can now load in the configuration data from a
properties file (or a Java Properties object).
For example, in your code, usually in the initialization stage:
_propConfig = new PropertiesConfigurator(
new File("logging.properties"));
_propConfig.configure();

That is it.
The example shown above reads in a properties file named logging.properties in the
current directory (you can fill in an absolute filepath), and triggers the configuration as
specified in the file. Please refer to the Javadoc API for various PropertiesConfigurator
constructors, including the option to load your own classloader.
Currently, there is not a centralized storage framework to hold the properties files. Each
application should refer to its own properties management directory. Should the tool cannot
locate the properties file (eg. wrong path, non-existing), an error message will be prompted,
and application should continue to run, in this case, without logging being switched on
(assuming all configuration is done through the properties file). Refer to Section 5.1 for
details on error handling concept.
4.6.1.2

Optional periodic reloading

There exists an optional command where you can specify a periodic value (in minutes) to
read and load the properties file again, to scan for any new configuration automatically.
This feature will be even more useful when a log console is available.
The API in doing this, e.g. reloading the properties file every 30 minutes:
_propConfig.setPeriodicity(30);

4.6.2 Syntax and semantics of properties file

The syntax rule and some semantics are clearly stated in the Javadoc API under the class
PropertiesConfigurator. In this section, we will elaborate on the explanation and examples on
the semantics in case you are not used to interpret the syntax rule. Bear in mind that this is like
Tutorial Logging & Tracing Mechanism in SAP
By: Infrastructure Management Team

transforming your configuration API into a list of commands of key-value-pair properties

represented in a properties file. So, it is a good idea to have the Javadoc API for reference while
you are reading this section. This can definitely speed up your understanding with the semantics of
the content.
4.6.2.1

A very simple sample content

As mentioned, it is listed in a key-value-pair format. The simplest form is:

<location>.attribute
= value
<category>.attribute
= value
(For your convenience, the attributes are bolded in the example to distinguish it from the
arbitrary <location> which is also delimited by a period .)
####Location "com.sap.foo"
com.sap.foo.severity
= WARNING
#### Now, assigning the log to the Location
com.sap.foo.logs
= ConsoleLog

Can you tell what this is doing? And can you map it back to the API?
This is equivalent to the following, where the configuration code is highlighted:
Location _loc = Location.getLocation(com.sap.foo);
_loc.setEffectiveSeverity(Severity.WARNING);
_loc.removeLogs();
//clean up existing attachments
//remember log assignment is additive?
_loc.addLog(new ConsoleLog());
_loc.fatalT(..);
_loc.infoT(.);

Now, with the use of the properties file, you can simply adjust the level of severity to
control the amount of output, or even redirect the messages into another destination(s).
You should notice the following sample values in the example above:
Attributes of a LogController: severity, logs
The value of severity: WARNING (constant value of class Severity, case-sensitive!)
The value of log: ConsoleLog (java class name of ConsoleLog, case-sensitive!)
Refer to the syntax rule described in the Javadoc API. Have you got a better sense of the
description? Can you follow the rule and map to the sample above?
4.6.2.2

Another sample content

The previous example is very straightforward. What if you need to do more configurations,
for example, assigning 2 output destinations for the location, replacing the default
configuration with your own configuration on the Log and/or on the Formattersuch as
assigning an XMLFormatter to a ConsoleLog (instead of using the default
TraceFormatter). In addition, you might want to reuse your own configuration object and
assign it to multiple other objects. It is no longer sufficient to use the classname
ConsoleLog directly.
You will have to use an identifier (~variable) to support these configurations. The notation
is not difficult: you denote the type of object, followed by the variable id in square bracket.
The 2 main types are:
log[<id>]
formatter[<id>]

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Then you can manipulate the configuration of the log or formatter object with its respective
attributes, before really using it. (Refer to next section 4.6.2.4 for a summary of possible
semantics.)
Imagine this is the configuration coding that we have originally. How will you represent
this in your properties file?
Location _loc = Location.getLocation(com.sap.foo);
_loc.setEffectiveSeverity(Severity.WARNING);
TraceFormatter _trFormatter =
new TraceFormatter(%s: %-30l [%t]: %m");
loc.addLog(new ConsoleLog(_trFormatter));
loc.addLog(new FileLog(C:\\temp\\myTrace.txt, _trFormatter));
_loc.fatalT(..);
_loc.infoT(.);

Once you understand the following representation, you are quite ready to deal with various
situations, with reference to the syntax rule indicated in the Javadoc API, in addition to the
summary tables listed in the next section.
com.sap.foo.severity
= WARNING
#### Set up a FileLog, for storing trace, with <id>: File
log[File]
= FileLog
log[File].pattern
= C:\\temp\\myTrace.txt
log[File].formatter
= formatter[TraceNoDate]
#### Set up a ConsoleLog, with <id>: Console
log[Console]
= ConsoleLog
log[Console].formatter
= formatter[TraceNoDate]
#### Set up a TraceFormatter, with <id>: TraceNoDate
#### and its pattern starts with the Severity level, and
consists no date/timestamp
formatter[TraceNoDate]
= TraceFormatter
formatter[TraceNoDate].pattern
= %s: %-30l [%t]: %m
com.sap.foo.logs

= log[Console], log[File]

In the very last line, you may realize that using a , makes multiple entries possible.
4.6.2.3

Support of periodic reloading

There is an intelligence to deal with log assignment (additive) with periodic reloading.
Technical details are not covered here but the general behavior will be explained and this
has something to do with the syntax: the use of an optional +, e.g.:
com.sap.foo.logs

= + log[Console]

A general rule of thumb: With the presence of a new log(s) configuration, without the +,
like the sample code in the previous 2 sections, the already assigned logs of the log
controller will be discarded, while with the +, the new logs will be added to the existing
list of logs.
4.6.2.3.1

Configuration unchanged
According to the rule description, that means on the other hand, when the configuration
remains unchanged (which is typical once a stable configuration has been defined), the
use of + is not significant at all. No further action will be done and the already

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

attached objects will still be attached. This avoids the loss of state of the attached
objects.
4.6.2.3.2

Configuration changed
The typical configuration updates include changing the log destination, adding log
destination and the modifying properties of an existing log.
The first 2 cases are straightforward, as described at the end of Section 4.4.6. Imagine
you have already assigned logX to a location. To change the destination to logY in the
next reloading, do the following in the configuration file:
com.sap.foo.logs

= logY

logX is removed.
To add to the existing logX:
com.sap.foo.logs

= + logY

Both logX and logY persist.

The last type of modification (changing properties) mainly applies to FileLog as it has
a number of attributes. A FileLog is identified by its pattern, therefore, a new pattern
simply means a brand new log and this behaves like the first 2 cases. The + is
significant.
If the changes are on the other attributes, like limit, cnt. The + is not significant,
no new FileLog will be created. The existing FileLog remains and will be modified
with the corresponding new attribute values. Recommendations for this siutation is not
to use the +, or use it for all occurrences of the log assignments.
4.6.2.4

Summary highlights

The syntax rule described in the Javadoc API helps you to validate your syntax, but perhaps
it is not too clear what attributes/values can be paired/assigned with/to what objects. Or
what are required, what are optional (refer to Appendix C for default values). The summary
below highlights the semantics. Note that the list is not exhaustive, a good trick is referring
to the Javadoc API for a reasonable semantics.
Class

Possible
Attribute

LogController
severity
effSeverity
minSeverity
maxSeverity
logs
localLogs
privateLogs
filters
bundleName
Log

FileLog

Formatter

severity
effSeverity
encoding
formatter
filters
pattern
limit
cnt
pattern

Reference to API

General for both Location, Category

<logController>.setEffectiveSeverity(int)
(same as severity)
<logController>.setMinimumSeverity(int)
<logController>.setMaximumSeverity(int)
<logController>.addLog(Log)
<logController>.addLocalLog(Log)
<logController>.addPrivateLog(Log)
<logController>.addFilter(Filter)
<logController>.setResourceBundleName(String)

General for all Log subclasses

<log>.setEffectiveSeverity(int)
(same as severity)
<log>.setEncoding(String)
<log>.setFormatter(Formatter)
<log>.addFilter(Filter)
new FileLog(String)
//Filename
new FileLog(String, int, int)
(same as limit)
<formatter>.setPattern(String) //message format

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Multiple
entries

Required

X
X
X
X

X
X

Table 4-5 Summary of configurable attributes of some classes

For primitive data type, you can easily assign an integer or string as an attribute value
(except Severity, see below). For attribute that takes a class, as listed above, such as Log,
Formatter, you will either use the classname directly (using default setting, see Section
4.6.2.1) or use the notation of a variable (see Section 4.6.2.2).
For the case of interface Filter, assign the fully qualified classname of the user-defined
class (make sure the class is already included in the classpath, and use an appropriate class
loader). As a matter of fact, this applies to user-defined classes for Log and Formatter as
well: use a full classname in case you are not using the standard classes provided by the
SAP API.
Attribute

Type

Possible Value

Description

severity, effSeverity,
minSeverity, maxSeverity

Integer

NONE

But constants defined in class Severity for your

convenience.
Refer to Javadoc of class Severity.
All are case-sensitive.

logs

(classname)

logs (variable)
formatter (classname)

formatter
Filters

(variable)
(classname)

Log

Formatter

Filter
(interface)

FATAL
ERROR
WARNING
INFO
PATH
DEBUG
ALL
FileLog
ConsoleLog
log[<id>]
TraceFormatter
ListFormatter
XMLFormatter
formatter[<id>]
com.sap.um.logg
ing.Filter

Again, case-sensitive
<id> can be any logical, meaningful name

<id> can be any logical, meaningful name

Actual implementation defined by users. The
arbitrary value shown on the left column
indicates a class named Filter is defined by
user under the package com.sap.um.logging.

If you understand the concept of relative severities (refer to Section 4.5.3) and would like to
use it, this can be expressed in the properties file as well. Again, use the notation of square
brackets.
Class

Attricute of Relative
severity

LogController
severity[<full name of relative
location or category>]
effSeverity[<full name of
relative location or category>]
minSeverity[<full name of
relative location or category>]
maxSeverity[<full name of
relative location or category>]

Reference to API

General for both Location, Category

<logController>.setEffectiveSeverity(Location, int)
<logController>.setEffectiveSeverity(Category, int)
(same as severity)
<logController>.setMinimumSeverity(Location, int)
<logController>.setMinimumSeverity(Category, int)
<logController>.setMaximumSeverity(Location, int)
<logController>.setMaximumSeverity(Category, int)

Eg. Category /System/Infrastructure/DB establishes a relative severity with location

com.sap.fooPackage.ClassX.
With coding API:
Location _loc =
Location.getLocation(com.sap.fooPackage.ClassX);
Category _cat =
Tutorial Logging & Tracing Mechanism in SAP
By: Infrastructure Management Team

Category.getCategory(/System/Infrastructure/Database);
_cat.setEffectiveSeverity(_loc, Severity.WARNING);

Equivalence in properties file:

System/Infrastructure/DB.severity[com.sap.fooPackage.ClassX]
= WARNING

4.6.2.5

Notes

What if the content of the properties file is invalid?

Normally, if the syntax or semantics is incorrect, or there is simply a typo, the
configuration will fail on that definition and prompt you an error message, explaining what
is going wrong and may take the default setting (see Appendix C). The tool will continue
to evaluate the subsequent configuration and output messages accordingly, based on
whatever successful configuration that has been made.
(The behavior results from the error handling mechanism designed in this tool. Refer to
Section 5.1)
Automatic lookup of default properties file?
No. There is no system property to define this, nor there is a default physical file location
specified for this. It is necessary to make one call in the program to load the properties.
This could be improved when a log console (section 6.2) is ready.
Can we define Location and/or Category object in the file?
First of all, users should not be stuck with the concept of creating a category or location
object and should not feel obliged to create one and maintain it. These are always
accessible by making calls like Category.getCategory() and
Location.getLocation() and the rest will be taken care of by the logging framework.
It is simply a good practice to use a static variable to store the handle for better
performance in case the source objects have to be accessed frequently. Thus, it is not
necessary to explicitly define/create these source objects in the configuration file in
advance; just use it directly in your program.
However, just a side note of interest, when you are specifying certain properties for a
source object in the configuration file, e.g. assigning severity or logs to a location or
category, the corresponding location or category will be implicitly instantiated and
correctly configured.
Does the order of configuration matter?
In general: NO.
Not for class variable definition. Consider the following, it is valid and correct.
com.sap.foo.logs
log[Console]
log[Console].formatter
log[File]
log[File].pattern
log[File].formatter
formatter[Trace]

formatter[Trace].pattern

= + log[Console], log[File]
= ConsoleLog
= ListFormatter
= FileLog
= %t/trace.log
= formatter[Trace]
= TraceFormatter

= %s: %24d: %m

Interestingly, order of severity assignment is not affected either. The severity inheritance is
controlled by the closest ancestor who has severity explicitly assigned.
com.sap.foo.fooChild.severity
= INFO
com.sap.foo.severity
= WARNING

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Imagine there are further descendants of com.sap.foo.fooChild, whom do not have

explicit severity assignment. With this sequence of configuration, what are their severities?
In this case, they actually inherit the severity from com.sap.foo.fooChild, not
com.sap.foo, so the answer is INFO.
For other straightforward type of assignments, the order does matter. Apparently, the latter
assignment will win. Easy example:
log[Console]
log[Console].formatter
log[Console].formatter

= ConsoleLog
= ListFormatter
= XMLFormatter

What will be the format of the output shown in the Console?

Refer to Section 4.6.2.3, the use of a + sign does not apply on log assignments specified
in the current file configuration content (it is affecting the already existing assigned logs,
such as from previous round of configuration in the case of periodic configuration). So, the
sequence still matters, and FileLog will win (assuming no previous configuration has been
done).
com.sap.foo.logs
com.sap.foo.logs

= + ConsoleLog
= + FileLog

If you want to assign multiple logs, simply use a comma as a delimiter:

com.sap.foo.logs

= + ConsoleLog, FileLog

5. Administrative issue
5.1

Error handling mechanism

Reason
The error handling mechanism should be done carefully in the logging tool. In any case, it should
not jeopardize the execution of the running application. Therefore, any error occurred within the
logging framework should be thrown and caught appropriately, be it a careless configuration error
or an unexpected system error.
Workaround
The approach to deal with this is to tolerate exception and to correct the error as best as we can
guess. The program will then continue to run. The exception is not lost; it is stored temporarily and
can be retrieved by the caller through the methods getException() and throwException().
In addition, we also make use of our own logging framework to log the exception message to the
internal dedicated category System/Logging, normally, with severity error and by default, output
to the Console (or you should know by now how to redirect the messages to a destination other
than the Console). Thus, users can get a better picture about the reason for certain not so standard
behavior while the application is still running smoothly. This proves to be very robust for runtime
exceptions such as java.lang.IllegalArgumentException,
java.lang.NullPointerException, .

Examples
A few potential mistakes are invalid severity definition, poor configuration, meaningless null
message string, wrong resource bundle information, etc. These can be rectified internally and a
message is logged to indicate the problem without terminating the program. There are several
Tutorial Logging & Tracing Mechanism in SAP
By: Infrastructure Management Team

ways to replace the erroneous value with the best-guess value: using default value (refer to
Appendix C) or estimating the closest value or simply ignoring it. For example:
Invalid configuration for FileLog for the pair of parameters limit and cnt. The default
setting for FileLog will be used, that is, limit and cnt will be both equal to zero.
For some reasons, if severity level has been out of the range of the defined Severity.MAX
and Severity.MIN, severity will be corrected to the closest limit (MIN or MAX)
If a typo occurs in defining a Log or Formatter, say in the configuration file, the value will
be ignored and nothing will be assigned
A message string or message code should not be a null string, but when this happens, a
constant <null> will be assigned.
When a message code cannot be resolved to a localized string, the value of message code
will be used directly
Note: exceptions generated from methods that are called in static field declarations and initializers
do lead to immediate program termination upon class loading. Therefore, errors like this can be
cited before software distribution.

5.2

Switching on/off the tool

Due to various reasons (eg. performance), logging is not turned on until requested by the
administrator at the time of application deployment. By and large, this logging tool should not
hinder your application performance of your application, but by default, logging is switched off
(This is achieved by the default severity level Severity.NONE assigned to all log controllers). This
aligns with the purpose of decoupling the process of inserting logging code and the task of
producing the actual logs/traces.
Eventually, this can be controlled through the log console. Currently, with the lack of interactive
user interface, this can still be achieved by the feature of reloading the configuration file
periodically. Initially, severity level is NONE (and this can be set globally in the configuration
file easily, making use of the hierarchical structure of the log controllers). With the specified
periodic value to reload the configuration file regularly (see Section 4.6.1.2), users can adjust the
severity level whenever they want to switch on the logging functions. This will automatically
picked up a the next reloading and configuration will become effective to start producing the
desired type and amount of messages.
(By the same token, you can always switch off the tool using the same technique.)

5.3

Compatibility with SUN JSR47

With the logging specification from SUN Microsystems, JSR47, there exists an implementation
java.util.logging package starting from JDK1.4, or some other third-party software which is
compliant to the specification. Although SAP logging API should be the standard tool to be used
by the SAP projects, there is a demand for a wrapper to the standard API.
We understand the standard API can be a prerequisite for certain groups because third-party
components are involved or that the SAP components to be run in a non-SAP environments,
therefore we are providing a wrapper so that users can continue to use this standard API in their
coding. It also provides the flexibility to stick with the standard JDK or to take advantage of the
SAP logging tool features at a later time.
For users who are interested in a compatible API and are still working in JDK1.3, we have a
wrapper to assist a smooth migration from 1.3 to 1.4. This wrapper is partially mapping the SUN

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

standard, mainly focusing on leaving your development code intact, with minimal penalty on
code modification. Howevre, currently, the loggingWrapper.jar is not publicly available.
Another wrapper implementation has been done for JDK1.4 developers to internally map to the
SAP logging API and objects. It is more versatile to use this wrapper for the standard API as you
have a choice to mingle the use of both the standard API and SAP API or simply stick with one
API. Coding and configuration can be reused readily with minimal penalty.
For more information about this, please refer to the Javadoc for this package:
com.sap.tc.loggingWrapper. Currently, the loggingWrapper.jar is not publicly available.

6. Future Development
6.1

Standard definition and rules

This will be an SAP-wide issue.
This involves, for example, predefined SAP Categories, location of output files, configuration
files
A rich set of standard Category should be available so that users can use readily. This includes the
typical logical areas like database, network, security, etc. These predefined categories will become
globally available and the standard for various groups at SAP. This will be very beneficial
because users do not have to randomly come up with a name, and need to worry about invalid
configuration. The maintenance will become more organized.
A good set of categories needs to be defined carefully to cover the common requirements of most
of the groups. Feedback and contribution from service group will be greatly helpful.
Other rules, such as standard locations of trace/log files, configuration files .etc should also be
determined. This will be decided upon by an expert group being put together at the moment.

6.2

Log console
This provides an interactive interface for users to easily manipulate the logging tool to suit their
needs, on both the logging behavior and the logging output. The current list of tasks is:
A message viewing tool
Options to allow selectively display trace and log messages
Interactive configuration (including starting and stopping the tool)
Support message translation at the time of viewing
This will be started by the InQMy group for the InQMy server.

6.3

Thread-specific tracing
Thread-specific and business-process-specific tracing is a well-known problem with server-based
systems. The execution of a single request might span across several servers (threads) and cause
problems. It is tricky to trace and analyze such request.

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Even if it is possible to tightly time a temporary modification of the global settings (within one
VM) of the Logging API, you usually get lots of messages from other requests that you are not
interested in. This hurts the system performance as well.
Thread-specific tracing allows the configuration specific to a single thread: above all setting a
lower severity for tracing in that single thread. In principle, this is done by putting these parts of
the configuration into a ThreadLocal. Based on these thread-specific settings, settings specific to a
request spanning several servers (threads) and even systems can be implemented (businessprocess-specific tracing).
To this end, the first thread in the chain packs its settings (which it gets from the front-end or a
corresponding properties file) into a transfer object. It then hands over this transfer object as an
argument to subsequent threads executing sub-requests. These threads in turn register the transfer
object with the Logging API, thus inheriting settings from callers. After finishing the request, the
Logging API is called again to forget about the settings specified via the transfer objects.

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Appendix A
Severity Levels
The following are the different levels of severity, in ascending order:
Severity
MIN
DEBUG

Value
0
100

PATH

200

INFO

300

WARNING

400

ERROR

500

FATAL

600

MAX

700

Details
Minimum restriction
For debugging purpose, with extensive and low level
information
For tracing the execution flow, e.g. used in the context of
entering and leaving a method, looping and branching
operations
Informational text, mostly for echoing what has been
performed
Application can recover from anomaly, and fulfill the
desired task, but need attention from developer/operator
Application can recover from error, but cannot fulfill
desired task due to the error
Application cannot recover from error, and the severe
situation causes fatal termination,
Maximum restriction

Table A-1 Definition of severity levels

Here are the other 2 useful severity keywords that will totally enable and disable tracing/logging:
Severity
ALL
NONE

Value
0
701

Details
Output messages of all severity
No messages will be logged

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Appendix B
Overview of Classes Hierarchy
Logging Manager

Log Controller

Category

Filter

Location

Filter

Log

Formatter

StreamLog

FileLog

ConsoleLog

ListFormatter

TraceFormatter

XMLFormatter

Figure B-1 Class diagram of the logging API

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Appendix C
Default behavior of certain objects
Class

Attribute

Default Value

Description

LogController

Severity
MinSeverity
MaxSeverity
<output logs>
<filters>
Severity
<filters>
Formatter
Formatter
Limit

Severity.NONE
Severity.ALL
Severity.NONE
<none>
<none>
Severity.ALL
<none>
TraceFormatter
ListFormatter
0

No output at all
No floor restriction: out all
No ceiling restriction: output nothing
No output at all

Log
ConsoleLog
FileLog

Count
TraceFormatter

Pattern

ListFormatter

Pattern

XMLFormatter

Pattern

0
%24d %-40l [%t] %s: %m

<!ELEMENT log (record*)>

<!ELEMENT table (id, time, source,
severity, location, thread, msg-type,
msg-code?, bundle?, msg-clear?,
args?)>
<!ELEMENT id (#PCDATA)>
<!ELEMENT time (#PCDATA)>
<!ELEMENT source (#PCDATA)>
<!ELEMENT severity (#PCDATA)>
<!ELEMENT location (#PCDATA)>
<!ELEMENT thread (#PCDATA)>
<!ELEMENT msg-type (#PCDATA)>
<!ELEMENT msg-code (#PCDATA)>
<!ELEMENT bundle (#PCDATA)>
<!ELEMENT msg-clear (#PCDATA)>
<!ELEMENT args (arg+)>
<!ELEMENT arg (#PCDATA)>

No further filtering
Do not further suppress messages
No further filtering

No file size limit. Count must == 0. No

rotating set of numbered files.
Logfile name generated without a
sequence number.
Jan 01, 2001 10:10:00 PM
com.sap.FooClass.fooMethod [main] Fatal: A
sample fatal message
#1.3#10.48.27.165:4A5AB2:E99D42D4F4:8000#Mon Jan 01 22:00:00 PDT
2001#com.sap.FooClass#com.sap.FooClass.fo
oMethod#main##0#0#Fatal##Plain###A
sample fatal message#
<record>
<id>10.48.27.165:4A5AB2:E99D2EDAFF:8000</id>
<time>Mon Jan 01 22:00:00 PDT
2001</time>
<source>com.sap.FooClass</source>
<location>com.sap.FooClass.fooMethod</locati
on>
<thread>main</thread>
<severity>Fatal</severity>
<msg-type>Plain</msg-type>
<msg-clear>A sample fatal message</msgclear>
</record>

Table C-1 Default attributes values

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team

Class

Property

Default Value

Description

Location

root name

Category

root name

FileLog

Filename,
with size limit
and count
Filename, if
not specified
by user
Message
displayed

<filename>.<ext>.cnt

Empty string. Full name should

be hierarchical, delimited by .
Similar to root directory. Full
name should be hierarchical,
delimited by /
Eg. trace.log.0, trace.log.1,

%t/trace.log

Eg. C:\temp\trace.log

Content of MsgClear
in any case

Message
displayed
Message
displayed
Output
Behavior by
relative

Optional

Either translated text in

MsgClear or text provided by
users explicitly in MsgClear
Both MsgCode & MsgClear
available, up to the viewer
Both MsgCode & MsgClear
available, up to the viewer
Relative A unaffected, and still
has default Severity.NONE and
no output destination.
Only when explicit configuration
is done on relative A, then
output will be generated from it

Trace

TraceFormatter

ListFormatter
XMLFormatter
Location or
Category with
relative A

Optional
Nothing will be output
by relative A

Table C-2 Highlights of default behavior

Tutorial Logging & Tracing Mechanism in SAP

By: Infrastructure Management Team