Chapter 1. Installing and Building the JBoss Server - JBoss Application Server 4.0.1 release1 Guide 英文版指南文档

Before installing and running the server, check your system to make sure you have a working JDK 1.4+ installation. The simplest way to do this is to execute the java -version command to ensure that the java executable is in your path, and that you are using Version 1.4 or higher. For example, running this command with a 1.4.1 JDK would produce version number like the following.

[tmp]$ java -version
java version "1.4.2_05"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_05-b04)
Java HotSpot(TM) Client VM (build 1.4.2_05-b04, mixed mode)

It does not matter where on your system you install JBoss. Note, however, that installing JBoss into a directory that has a name containing spaces causes problems in some situations with Sun-based VMs. This is caused by bugs with file URLs not correctly escaping the spaces in the resulting URL. There is no requirement for root access to run JBoss on UNIX/Linux systems because none of the default ports are within the 0-1023 privileged port range.

After you have the binary archive you want to install, use the JDK jar tool (or any other ZIP extraction tool) to extract the jboss-4.0.1.zip archive contents into a location of your choice. The jboss-4.0.1.tgz archive is a gzipped tar file that requires a gnutar compatible tar which can handle the long pathnames in the archive. The default tar binaries on Solaris and OS X do not currently support the long pathnames. The extraction process will create a jboss-4.0.1 directory. The following section explores the contents of this directory.

1.1.2.1. Directory Structure

As mentioned above, installing the JBoss distribution creates a jboss-4.0.1 directory which contains server start scripts, JARs, server configuration sets and working directories. You do need to know your way around the distribution layout to locate JARs for compilation, updating configurations, deploying your code, etc. Figure 1.1, “The view of the JBoss server installation directory structure with the default server configuration file set expanded and overridable locations identified” illustrates the installation directory of the JBoss server.

Figure 1.1. The view of the JBoss server installation directory structure with the default server configuration file set expanded and overridable locations identified

Throughout the book we will refer to the top-level jboss-4.0.1 directory as the JBOSS_DIST directory. In Figure 1.1, “The view of the JBoss server installation directory structure with the default server configuration file set expanded and overridable locations identified”, the default server configuration file set is shown expanded. It contains a number of subdirectories: conf, data, deploy, lib, log and tmp. In a clean installation, only the conf, deploy and lib directories will exist. The purpose of each of these directories are discussed in Table 1.1, “The JBoss directory structure”. In this table, the ServerConfig Property column refers to the org.jboss.system.server.ServerConfig interface constant and its corresponding system property string. The ServerConfig constant names and corresponding system property name are displayed in the blue text in Figure 1.1, “The view of the JBoss server installation directory structure with the default server configuration file set expanded and overridable locations identified”. The XXX_URL names correspond to locations that can be specified using a URL to access remote locations, for example, HTTP URLs against a web server. You can use the properties listed in the following table to override the layout of a JBoss distribution

Table 1.1. The JBoss directory structure

Directory	Description	ServerConfig Property
bin	All the entry point JARs and start scripts included with the JBoss distribution are located in the bin directory.
client	JARs required for clients are located in the client directory. A typical client requires the following: jbossall-client.jar concurrent.jar log4j.jar
server	The JBoss server configuration sets are located under the server directory. The default server configuration set is the server/default set. JBoss ships with minimal, default and all configuration sets. The subdirectories and key configuration files contained in the default configuration set are discussed in more detail in Section 1.1.3, “The Default Server Configuration File Set”	SERVER_BASE_DIR ="jboss.server.base.dir"SERVER_BASE_URL ="jboss.server.base.url"
lib	The lib directory contains startup JARs used by JBoss. Do not place your own libraries in this directory.	LIBRARY_URL ="jboss.lib.url"
conf	The conf directory contains the bootstrap descriptor, jboss-service.xml by default, file for a given server configuration. This defines the core services that are fixed for the lifetime of the server.	SERVER_CONFIG_URL ="jboss.server.config.url"
data	The data directory is a location available for use by services that want to store content in the file system.	SERVER_DATA_DIR ="jboss.server.data.dir"
deploy	The deploy directory is the default location the hot deployment service looks to for dynamic deployment content. This may be overridden through the URLDeploymentScanner URLs attribute.
lib	The lib directory is the default location referred to the by bootstrap descriptor. All JARs in this directory are loaded into the shared classpath.	SERVER_LIBRARY_URL ="jboss.server.lib.url"
log	The log directory is the default directory into which the bootstrap logging service places its logs. This may be overridden through the conf/log4j.xml configuration file.	none
tmp	The tmp directory is the location to which deployments are copied for local use.	SERVER_TEMP_DIR="jboss.server.temp.dir"

The JBOSS_DIST/server directory contains one or more configuration file sets. The default JBoss configuration file set is located in the JBOSS_DIST/server/default directory. JBoss allows you to add more than one configuration set so a server can easily be run using alternate configurations. Creating a new configuration file set typically starts with copying the default file set into a new directory name and then modifying the configuration files as desired. Figure 1.2, “An expanded view of the default server configuration file set conf and deploy directories” below shows the contents of the default configuration file set.

Figure 1.2. An expanded view of the default server configuration file set conf and deploy directories

The JBoss source is hosted at SourceForge, a great Open Source community service provided by VA Linux Systems. With over 88,000 Open Source projects and nearly 950,000 registered users, SourceForge.net is the largest Open Source hosting service available. Many of tche top Open Source projects have moved their development to the sourcesorge.net site. The services offered by SourceForge include hosting of project CVS repositories and a web interface for project management that includes bug tracking, release management, mailing lists and more. Best of all, these services are free to all Open Source developers. For additional details and to browse the plethora of projects, see the SourceForge home page. (http://sourceforge.net/).

CVS (Concurrent Versions System) is an Open Source version control system that is used pervasively throughout the Open Source community. CVS is a source control or revision control tool designed to keep track of source changes made by groups of developers who are working on the same files. CVS enables developers to stay in sync with each other as each individual chooses.

The JBoss project's SourceForge CVS repository can be accessed through anonymous (pserver) CVS with the following instruction set. The module you want to check out must be specified as the modulename. When prompted for a password for anonymous, simply press the Enter key. The general syntax of the command line version of CVS for anonymous access to the JBoss repositories is:

cvs -d:pserver:anonymous@cvs.sourceforge.net:/cvsroot/jboss login
cvs -z3 -d:pserver:anonymous@cvs.sourceforge.net:/cvsroot/jboss co modulename

The first command logs into JBoss CVS repository as an anonymous user. This command only needs to be performed once for each machine on which you use CVS because the login information will be saved in your HOME/.cvspass file or equivalent for your system. The second command checks out a copy of the modulename source code into the directory from which you run the cvs command. To avoid having to type the long cvs command line each time, you can set up a CVSROOT environment variable with the value :pserver:anonymous@cvs.jboss.sourceforge.net:/cvsroot/jboss and then use the following abbreviated versions of the previous commands:

cvs login
cvs -z3 co modulename

The name of the JBoss module alias you use depends on the version of JBoss you want. For the 4.0 branch the module name is jboss-4.0, for the 3.2 branch it is jboss-3.2, and in general, for branch x.y the module name is jboss-x.y. To checkout the HEAD revision of jboss to obtain the latest code on the main branch you would use jboss-head as the module name. Releases of JBoss are tagged with the pattern JBoss_X_Y_Z where X is the major version, Y is the minor version and Z is the patch version. Release branches of JBoss are tagged with the pattern Branch_X_Y. Some checkout examples are:

cvs co -r Branch_4_0 jboss-4.0 # Checkout the current 3.0 branch code
cvs co -r JBoss_4_0_1 jboss-4.0 # Checkout the 4.0.1 release version code
cvs co -r JBoss_3_2_6 jboss-3.2 # Checkout the 3.2.6 release version code
cvs co jboss-head # Checkout the curent HEAD branch code

The command line version of the CVS program is freely available for nearly every platform, and is included by default on most Linux and UNIX distributions. A good port of CVS as well as numerous other UNIX programs for Win32 platforms is available from Cygwin at http://sources.redhat.com/cygwin/. The syntax of the command line version of CVS will be examined because this is common across all platforms.

For complete documentation on CVS, check out the CVS home page at http://www.cvshome.org/.

Every JBoss release includes a source archive that contains everything needed to build the release and is available from the files section of the JBoss project site here: http://sourceforge.net/projects/jboss/. The source directory structure matches that of the CVS source tree described below so once you have the source distribution you can build the release by following the instructions given in the next section, beginning with the instructions after the step to obtain the jboss-4.0 source tree.

This section will guide you through the task of building a JBoss distribution from the CVS source code. To start, create a directory into which you want to download the CVS source tree, and move into the newly created directory. This directory is referred to as the CVS_WD directory for CVS working directory. The example build in this book will check out code into a /tmp/jboss directory on a Linux system. Next, obtain the 4.0.1 version of the source code as shown here:

[tmp]$ mkdir jboss
[tmp]$ cd jboss
[jboss]$ export CVSROOT=:pserver:anonymous@cvs.sourceforge.net:/cvsroot/jboss
[jboss]$ cvs co -r JBoss_4_0_1 jboss-4.0
cvs checkout: Updating aop
U aop/.classpath
U aop/.cvsignore
U aop/.project
U aop/RELEASE_NOTES.txt
...

The resulting jboss-4.0 directory structure contains all of the CVS modules required to build the server. To perform the build, cd to the jboss-all/build directory and execute the build.sh or build.bat file as appropriate for you OS. You will need to set the JAVA_HOME environment variable to the location of the JDK you wish to use for compilation, and the JDK bin directory should be on your path.

[jboss]$ cd jboss-4.0/build/
[build]$ ./build.sh
Searching for build.xml ...
Buildfile: /tmp/jboss/jboss-4.0/build/build.xml
...
 
BUILD SUCCESSFUL
Total time: 2 minutes 41 seconds
                

Note that if you see an "Failed to launch JJTree" error do not have the JAVA_HOME/bin directory in your PATH required for the JavaCC JJTree Ant task.

The build process is driven by an Ant based configuration. The main Ant build script is the build.xml file located in the jboss-4.0/build directory. This script uses a number of custom Ant tasks masked as buildmagic constructs. The purpose of the main build.xml file is to compile the various module directories under jboss-4.0 and then to integrate their output to produce the binary release. The binary release structure is found under the jboss-4.0/build/output directory. The example above used the build.sh script to kickoff the build process. This is just a wrapper the launches the ant binary included in the distribution. You can use Ant directly to build if your environment is configured run it from the command line.

The top-level directory structure under the jboss-4.0 source tree is illustrated in Figure 1.3, “The testsuite CVS module directory structure”, the CVS source tree top-level directories. Table 1.2, “Descriptions of the top-level directories of the JBoss CVS source tree.” gives the primary purpose of each of the top-level directories.

More advanced testing of the JBoss installation and builds can be done using the JBoss testsuite. The JBossTest suite is a collection of client oriented unit tests of the JBoss server application. It is an Ant based package that uses the JUnit (http://www.junit.org) unit test framework. The JBossTest suite is used as a QA benchmark by the development team to help test new functionality and prevent introduction of bugs. It is run on a nightly basis and the results are posted to the development mailing list for all to see.

The unit tests are run using Ant and the source for the tests are contained in the jboss-4.0/testsuite directory of the source tree. The structure of the testsuite CVS module is illustrated in Figure 1.3, “The testsuite CVS module directory structure”.

Figure 1.3. The testsuite CVS module directory structure

The two main source branches are src/main and src/resources. The src/main tree contains the Java source code for the unit tests. The src/resources tree contains resource files like deployment descriptors, JAR manifest, web content, etc. The root package of every unit test is org.jboss.test. The typical structure below each specific unit test subpackage (for example, security) consists of a test package that contains the unit test classes. The test subpackage is a required naming convention as this is the only directory searched for unit tests by the Ant build scripts. If the tests involves EJBs then the convention is to include an interfaces and ejb subpackage for these components. The unit tests themselves need to follow a naming convention for the class file. The unit test class must be named XXXUnitTest.java, where XXX is either the class being tested or the name of the functionality being tested.

To run the unit tests use the build scripts located in the testsuite directory. The key targets in the build.xml file include:

On completion of a test the testsuite/output/reports directory will contain one or more XML files that represent the individual JUnit test runs. The tests-report target collates these into an HTML report located in the html subdirectory along with a text report located in the text subdirectory. Figure 1.4, “An example testsuite run report status HTML view as generated by the testsuite” shows an example of the HTML report for a run of the test suite against the JBoss 4.0.1 release.

Figure 1.4. An example testsuite run report status HTML view as generated by the testsuite

You can find the results of the testsuite in the JBoss distribution in under the JBOSS_DIST/docs/tests directory.