Logging with log4j

With the log4j.xml configuration file, you can do the following:

• Categorize and prioritize log statements according to your preferences, which allows you to specify the degree of significance a statement must have to be logged to the log repository.
• Allocate log statement destination; you can specify unique destinations for different specified categories.
• Create custom formats for log output (or rely on provided default format).

In the absence of a configuration file (such as log4j.xml or log4j.properties), Viewpoint saves logs to the /tmp directory, or to the Tomcat temporary directory ($CATALINA_HOME/temp/)

The log4j.xml file is checked every 10 seconds for changes. Therefore, after any changes are made to the configuration, it is not necessary to restart Viewpoint, but it may take up to 10 seconds for the changes to be reflected in the logs.

The following is an example of the Viewpoint default logging properties:

<!DOCTYPE log4j:configuration SYSTEM "log4j.dtd" >
<log4j:configuration>

	<appender name="file" class="org.apache.log4j.RollingFileAppender">
		<param name="maxFileSize" value="50MB" />
		<param name="maxBackupIndex" value="5" />
		<param name="File" value="${VIEWPOINT_HOME}/log/viewpoint.log" />
		<layout class="org.apache.log4j.PatternLayout">
			<param name="ConversionPattern" value="%d{dd MMM yyyy HH:mm:ss,SSS} %5p %c{1}:%L - %m%n" />
		</layout>
	</appender>

	<!-- Logger for Viewpoint server code -->
	<logger name="com.cri" additivity="false">
		<level value="info"></level>
		<appender-ref ref="file"></appender-ref>
	</logger>

	<!-- Logger for Viewpoint security -->
	<logger name="com.cri.security" additivity="false">
		<level value="debug"></level>
		<appender-ref ref="file"></appender-ref>
	</logger>

	<!-- Logger for external Java API classes -->
	<logger name="com.ace" additivity="false">
		<level value="info"></level>
		<appender-ref ref="file"></appender-ref>
	</logger>

	<!-- Logger for low level internal Java API classes -->
	<logger name="com.moab" additivity="false">
		<level value="info"></level>
		<appender-ref ref="file"></appender-ref>
	</logger>

	<!-- Logging for everything but the Viewpoint code and Moab Java API. This includes 
		hibernate and its associated connection pool -->
	<root>
		<priority value="info"></priority>
		<appender-ref ref="file"></appender-ref>
	</root>

</log4j:configuration>

Appender

An output destination is called an appender. Log4j allows logging requests to print to multiple destinations. Multiple appenders can be attached to a logger. The <appender> tag describes the logger type, which is a file appender. This appends the log to the specified file in the appender. The Java System property VIEWPOINT_HOME is used. This is set to the environment variable of $VIEWPOINT_HOME upon web application startup. The default path to the files that contain the logs are contained at /opt/viewpoint/log/viewpoint.log.

Logger

The <logger> tags describe the different log levels that can be changed according to the Java packages of the source code. The list that follows describes the packages used in the default configuration. Note also that each logger tag has an associated value that indicates the verbosity level. More information on verbosity levels is available in the next section.

• com.cri - This is the general package for all Viewpoint server code. This will log all server interaction with the Viewpoint web server.
• com.cri.security - This is the logger for all security packages. This is important for authentication and authorization testing.
• com.ace - This is the logger for external Java API classes. This includes Java code that has been validated for external customer consumption. Classes include MoabJob, MoabNode, Reservation, and so forth. See the Moab Java API for more information.
• com.moab - This is the logger for internal Java API classes. Classes include VPCs, low-level Moab consumption, etc. This is the logger that shows what Moab commands are sent and the associated output.
• Root Logger - This is the logger for everything other than the specified loggers. This is a catch-all for logging of the Tomcat server, Hibernate, the database connection pool, and other third party libraries. Unless the packages are explicitly specified in a logger like the previous examples, the logging will fall into the root logger.

Setting Logging Verbosity Levels

Each logger tag requires an associated level of verbosity. The levels are cumulative, meaning that each level also logs the messages logged by levels hierarchically ordered below it. For example, if an administrator configures a logging level of info, method calls to logger.info(), logger.warn(), logger.error(), and logger.fatal() are also logged. The following table offers a brief description of the various log4j verbosity levels:

Verbosity Levels

Level	Definition
trace	All data.
debug	Normal operations.
info	Significant system events such as application initialization, logging initialization, and occurrence of rare global events.
warn	Recoverable exceptions or unexpected state.
error	Unrecoverable request exceptions or recoverable system errors.
fatal	Indicates occurrence of an unrecoverable system error resulting in a crash of the application.

Example

In Viewpoint, at application startup, a dynamically loaded class authenticates HTTP requests. However, if the class does not exist in the classpath, or if there is some other type of exception in creating an instance of the class, a message is logged via logger.fatal() to explain why the application failed.