------
 Installing Continuum Standalone
 ------
 Emmanuel Venisse
 ------
 2008-09-01
 ------

Installing Continuum Standalone

~~%{toc|section=0}

* {Basics & Fundamentals}

    * Download the standalone version from the {{{http://continuum.apache.org/download.html} Download page}}

    * Extract the file
    
    * Set a JAVA_HOME environment variable which use a jdk >= 1.5 

** {Defining JNDI Resources}

*** {Mail server configuration}

    Before to start Continuum, you must configure your SMTP configuration for mail notification. The configuration to do is in <<<$CONTINUUM_HOME/conf/jetty.xml>>>:

%{snippet|id=mail|url=http://svn.apache.org/repos/asf/continuum/trunk/continuum-jetty/src/main/conf/jetty.xml}

*** {Databases configuration}

    By default, Continuum use an embedded {{{http://db.apache.org/derby}Derby}} database. If you want to use an other database, you can modify the JNDI
    configuration in <<<$CONTINUUM_HOME/conf/jetty.xml>>>:

%{snippet|id=datasources|url=http://svn.apache.org/repos/asf/continuum/trunk/continuum-jetty/src/main/conf/jetty.xml}

* {Installing as a Windows Service}

    * Go to <<<$CONTINUUM_HOME/bin/>>> and run the following command:

------------------
continuum.bat install
------------------

    * Edit the <<<'Apache Continuum'>>> service

        * To see the services that are on your computer go to Start|Run and enter <<<'services.msc'>>>.

    * Select the <<<'Startup Type'>>>

    * Go to the <<<'Log On'>>> tab and select a real user. A real user is required because you'll need a home directory for maven repository and some other things

    * Validate your changes

* {Installing as a Linux Service}

    Since the Continuum linux script bin/linux/run.sh understands the same arguments as linux boot scripts, there is no need to write a particular
    startup script to add Continuum to the linux boot process. All you need to do, as root, is:
    
** {Basic script in /etc/init.d}

    * Create a <<<'continuum'>>> file under <<</etc/init.d/>>> with the following content (replacing <<<continuum_user>>> with the name of an account you have already created):

------------------
#!/bin/sh

CONTINUUM_HOME=/opt/continuum-1.2
su - continuum_user -c "$CONTINUUM_HOME/bin/continuum console $@ &"
------------------

** {In a Debian-based system}

------------------
ln -s /usr/local/continuum-[VERSION]/bin/linux/run.sh /etc/init.d/continuum
------------------

    At this point you have Continuum ready to be symlinked from different runlevels. This might sound a bit esoteric, but it is not, you will find these words
    very fast as soon as you start reading about the init process. Fortunately, Debian GNU/Linux comes with a very handy utility to create this links, just run as root:

------------------
update-rc.d -n continuum defaults 80
------------------

    If you run this command, you will see something like this:

------------------
 Adding system startup for /etc/init.d/continuum ...
   /etc/rc0.d/K80continuum -> ../init.d/continuum
   /etc/rc1.d/K80continuum -> ../init.d/continuum
   /etc/rc6.d/K80continuum -> ../init.d/continuum
   /etc/rc2.d/S80continuum -> ../init.d/continuum
   /etc/rc3.d/S80continuum -> ../init.d/continuum
   /etc/rc4.d/S80continuum -> ../init.d/continuum
   /etc/rc5.d/S80continuum -> ../init.d/continuum
------------------

    What you see is the symlinks that would be created. The above command didn't do anything because of the -n  switch, remove it to get the real links created.

** {In a RedHat-based system}

    Configuring Continuum in a RedHat-based system (like Fedora Core) is slightly different: Instead of running update-rc.d, you need to add a new
    service using chkconfig. And in order to add Continuum to chkconfig, it is necessary to add some comments to the /etc/rc.d/init.d/continuum script
    and run a couple of commands; these tasks are automatically executed by running the chkconfig_install.sh script (note that _continuum_user_ needs to be
    replaced by the name of an account you have already created):

------------------
#! /bin/sh
#
# chkconfig_install.sh - install Continuum on a chkconfig-bases system
# 
# Author: Felipe Leme <felipeal at apache.org>
#

# figure out what's Continuum's directory
CONTINUUM_HOME=`dirname $0`
cd ${CONTINUUM_HOME}
CONTINUUM_HOME=`pwd`

INITD_SCRIPT=/etc/rc.d/init.d/continuum

if [ -f ${INITD_SCRIPT} ]
then
  echo "File ${INITD_SCRIPT} already exists. Please remove it and try again."
  exit 1
fi

echo "Creating file ${INITD_SCRIPT}"  
cat >> ${INITD_SCRIPT} <<EOF
#! /bin/sh
# chkconfig: 345 90 10
# description: Apache Continuum server

# uncoment to set JAVA_HOME as the value present when Continuum installed
#export JAVA_HOME=${JAVA_HOME}

if [ -z "\${JAVA_HOME}" ]
then
  echo "Cannot manage Continuum without variable JAVA_HOME set"
  echo "  (try to set it on file ${INITD_SCRIPT})"
  exit 1
fi
# run Continuum as root
cd ${CONTINUUM_HOME}
./run.sh \$*
# run Continuum as user _continuum_user_
#su - _continuum_user_ -c "cd ${CONTINUUM_HOME}; ./run.sh \$*"
EOF
chmod +x ${INITD_SCRIPT}

echo "Adding Continuum to chkconfig"  
chkconfig --add continuum

echo "Enabling Continuum on chkconfig"  
chkconfig continuum on
echo "Continuum set to start on run levels 3, 4 and 5."
echo "To start continuum now, run 'service continuum start'"

------------------