Easy file transfer

• SCP
  scp /home/xxx/middleware/jon-server/test.txt root@192.168.100.150:/app/jon/bin

• KDE - Dolphin

fish://root@192.168.100.150
fish://root@192.168.100.150/app/jon

Get a Red Hat red hat

http://redhat.brandfuelstores.com/index.php?main_page=index&cPath=2_17

Writing start scripts for services in Fedora linux

Howto manage your services and programs within Fedora

1. Make your shell script executable

All shell scripts in /usr/bin/ or /sbin/ can be executed anywhere within the command line

Add a softlink into /sbin or /usr/bin

$ cd /sbin

$ ln -s /app/jon/bin/rhq-server.sh jon

$ jon

1.5. Start your shell script at startup by modifying

Example:

$ sudo vi /etc/rc.d/rc.local

add:
killall -9 knetworkmanager & qdbus org.kde.kded /kded unloadModule networkmanagement && nm-applet &

2. Start your program easily via shell script

If you have a shell that takes the standard arguments : start / stop / restart

you can it into the repository /etc/init.d/.

All scripts in /etc/init.d/ can be called with

$ service my_script start

For convenience use a softlink instead of copying it into /etc/init.d:

$ cd /etc/init.d

$ ln -s /app/jon/bin/rhq-server.sh jon

$ service jon start

3. Service registry within Fedora (the startup services)

List all available services (/etc/init.d)

$ chkconfig --list

Start / stop the ssh daemon service

$ chkconfig sshd start (or stop)

Enable a service

$ chkconfig --add ssh

$ chkconfig sshd on

Disable a service

$ chkconfig sshd off

$ chkconfig sshd --del

If you you prefer a graphical interface for the service registry use

$ system-config-services

4. Making an executable 'servicable'

Create a file in /etc/init.d, here an example: shutter 
#!/bin/sh
#
# shutter This script starts and stops the shutter daemon
#
# chkconfig: - 78 30
# processname: shutter
# description: shutter is a daemon process

get_shutter_pid () {
   SHUTTERPID=$( ps aux --cols 1024 | grep "perl" | grep "/usr/bin/shutter" | awk '{ print $2 }' | head -n 1 )
}

PIDFILE=/var/run/shutter.pid

# By default it's all good
RETVAL=0

BASE="/usr/bin"

# See how we were called.
case "$1" in
  start)
    # Start daemon.
    echo -n $"Starting shutter: "
    export DISPLAY=:0
    nohup ${BASE}/shutter > /dev/null &
        sleep 1
        get_shutter_pid
        echo -n ${SHUTTERPID} > ${PIDFILE}
        echo "Start Shutter : pid = ${SHUTTERPID}"
        ;;
  stop)
        # Stop daemons.
        get_shutter_pid
        if ! [ -z "${SHUTTERPID}" ] ; then
           echo "Stop Shutter : pid = ${SHUTTERPID}"
           kill "${SHUTTERPID}"
        fi
        ;;
  restart|force-reload)
        $0 stop
    sleep 8
        $0 start
        ;;
  status)
        get_shutter_pid
        if [ -z "${SHUTTERPID}" ] ; then
            echo "Shutter is stopped"
            exit 1
         else
            echo "Shutter (pid ${SHUTTERPID}) is being executed..."
            exit 0
         fi
    ;;
  *)
    echo "Usage: $0 {start|stop|restart|status}"
    RETVAL=1
    ;;
esac

exit $RETVAL

Howto use JBoss Operation Network (JON) for your provisioning

Post summary:
This post will illustrate how you can set up a Red Hat middleware provisioning environment using Linux (Fedora), KVM (virtualisation), JBoss EAP (application server) and JBoss Operations Network (supervision) and Apache httpd (mod_cluster).
A use case could be : JON detects that the an application deployed on EAP can not handle the charge any more and triggers the creation of a new virtual machine that contains a new JBoss EAP. Via a bundle the application is deployed in EAP who joins the cluster.

Main post :
Here I more or less dumped the technical steps to follow without explaining step by steps (as it's a long post use ctrl+F to be happy)

On host:

• sudo service sshd start

Connect to JON VM:

• cd /app/script
• ./initSshId.sh (removing the file /etc/.ssh/know_hosts might be necessary)
• Creation of a soft link for your script from the JON/bin directory to the script directory. As you can see in the next screen shot, this will allow JON to discover automatically those scripts (without soft link you would need a plugin)
  cd  /app/jon/jon-server/bin
  ln -s /app/script/virsh-start.sh virsh-start.sh

        

KDE vs Gnome

Summary:
Having the Gnome desktop you want to try the KDE desktop  (or vice versa)

Main post:

• Download the kde packages
  sudo yum install kdm (this is only the startup manager)
  sudo yum install kde
• Setting KDM as startup manager
  su -
  echo "DISPLAYMANAGER=KDE" > /etc/sysconfig/desktop
• Choose your desktop environment
  switchdesk (switch from Gnome to KDE desk within your session)
• Kill your XServer from console
  in Gnome:
  cd /etc/init.d/
  sudo gdm-stop
  in KDE:
  init 3
• Start your XServer
  xstart

Disable SELinux on Fedora 13

• Edit the following file:
  sudo vi /etc/selinux/config
• Replace
  SELINUX=enforcing with SELINUX=disabled

• Stop X server
  telinit 3  (or within gnome gdm restart)

• Start X server
  telinit 5

My useful Linux commands

Some Linux commands I find useful:

Replace a string in all files recursively usind sed ( replaces repository.jboss.com with anonsvn.jboss.org/repos/repository.jboss.org)

grep -rl repository.jboss.com . | xargs sed -i 's/repository.jboss.com/anonsvn.jboss.org\/repos\/repository.jboss.org/'

Add launchers (*.desktop, icon - comment - name) here
/usr/share/applications

Find a file on the file system

updatedb

locate test.xml

Find out which kernel you are using
uname -r

Find string "proxy" in all files
sudo find . -type f -name "*" -exec grep -l "proxy" {} \;

find . -name "*" -exec grep "proxy" {} \;      (donne tous les lignes avec Test_1)

find . -name "*" -exec grep -l "proxy" {} \ ; (donne tous les fichiers avec Test)

grep -r 'proxy' *

Making a script executable
chmod +x

Finding all my files with string "myfile" including soft link references directories
find . -user me -name *myfile* -follow


Displaying directory structure
ls -rtl bbb (-R subdirectories)


Monitoring changes within a file
tail -100 fichier
tail -f file (monitoring)

grep -F ‘abc efg’ (Leerstelle = OR)
grep -i -n (line number) testChar file.txt

grep « VALOTRACE= » `(Alt + Shift + 7) find .`

cat fichier | grep blub
Displays full path of file being executed
which java

grep valopurgesolde.x communs/ces/x (pour voir qu’est-ce que le processus fait)
kill -9 -1 (killt das zuletzt angemeldete Konto)
cd - (go to root)
du -sk . (Wie viel Speicherplatz belegt)
df -k (kilo) . (combien il reste sur le filesystem)
ps -u valqua11  (kill -9 #1 #2)

ln -s link name (ln -s ${HOME}/brackmann roli)
Displaying all exports of the shell
typeset
Process keeps on running in background even when the executing shell is killed
nohup your_command

Gets the 4th columns of the server.log and deletes all the doubles

cat server.log | awk '{ print $4; }' | sort -u > server4thcolumndistinct.log


Deletes all temporary files followed by '~'
find ./ -name '*~' | xargs rm

Classloading JBoss

Title : Web app is using embedded jars that exist already within JBoss

Summary:
First of all w
JBoss class loading
A college of mine has written a good article on class loading :
The translated Noel Rocher

Main post
    jboss-app.xml         <?xml version="1.0" encoding="utf-8"?>
        <jboss-app>
            <loader-repository>
            TestClassloaderJBossLR:loader=TestClassloaderJBoss.ear
            <loader-repository-config>java2ParentDelegation=true</loader-repository-config>
            </loader-repository>
        </jboss-app>   
    jboss-web.xml

        <?xml version="1.0" encoding="utf-8"?>
        <jboss-web>
            <!-- Mettre toute l'application soit dans le meme classloader (J2EE) -->
            <class-loading java2ClassLoadingCompliance="true"/>
        </jboss-web>
 
Ever since version 4.0.2, JBoss is configured according to the J2EE specification, meaning that the default isolation of applications is the research in priority classes in the ClassLoader of the application.

Specifying in the 2-descriptors jboss and jboss-web.xml app.xml attributes and java2ClassLoadingCompliance java2ParentDelegation = true = "true, you change this default behavior and return to as it was in JBoss 3.2, namely the use of a unified ClassLoader taking priority classes loaded first, so in your case those server applications.
fsfdf

Webservices within JBoss

Summary:
Web services give you the possibility of loosely coupling your different IT-systems. At least on the paper it seems that you can interchange either the client or server easily.
Only after having changed either part for the first time you might realize that a new Web service implementation on one side means redeveloping also the other.

Main post:
When working with the Webservice stack of JBoss you will have to check on the version you are using, as there has been some changes over the years :

http://community.jboss.org/wiki/WebServiceStacks

Name	File	JBoss Version	Status	Supported API	Based on
JbossWS-3.x	JbossWS Native 3.1.2.SP3-patch-02 jbossws.sar	5.0+		JEE 6	JAX-WS
	Apache CXF (not shipped with EAP) 3.1.2.SP3	5.0+	Supported	JEE 6 (1)	JAX-WS
	GlassFish Metro (not shipped with EAP)	5.0+		JEE 6	JAX-WS
JBossWS-2.x sources binaries	jbossws.sar (2.0.1.SP2_CPxx)	EAP 4.3 CPxx	???	JEE 5	JAX-WS
JBossWS-1.x	jbossws.sar (1.2.1.GA_CPxx+)	EAP 4.2 CPxx	???	JEE 5	JAX-WS (2)
JBossWS4EE	jboss-ws4ee.sar	< 4.0.4	Maintenance Only	J2EE 1.4	Axis
JBoss.NET	jboss-net.sar	3.2.x	End of Life	Proprietary	Axis

(1) JDK 1.6 incorporates the JAXB reference implementation. However, it incorporates an old version of the RI. CXF does not support this version. As of 1.6_04, this is easy to deal with: you must put the versions of JAXB RI (the 'impl' and 'xjc' jars) that we include with CXF in your classpath. As of this writing, these are version 2.1.6.
(2) Yes, JAXWS ... are supported all together starting from JBossWS-1.2.0.GA. Link

This table was build upon http://community.jboss.org/wiki/WebServiceStacks, http://community.jboss.org/wiki/JBossWS, http://community.jboss.org/wiki/JBossWS-Installation and notes from Alessio Soldano

Howto log within JBoss

Main point of entry:
http://docs.jboss.org/process-guide/en/html/logging.html

Linux /boot partition problem (upgrading Fedora Linux)

Summary:
/boot partition too small when trying to upgrade Fedora 10 to 11, 11 to 12, 12 to 13, 13 to 14

Main article:
When not choosing a big enough /boot partition when setting up you Linux system you might get problems when wanting to upgrade to the next major Fedora release.
Fedora (same with Ubuntu) will tell you that space on /boot is missing.
Don't try lvm (Logic Volume Managment) it won't help you!

You have the following possibilties
1. Delete old kernel updates
package-cleanup --oldkernels
2. Empty trash
du -sh .Trash-0
3.
tune2fs -m 0/dev/sda1

Howto check your J2EE application

http://j-francois.blogspot.com/2010/04/check-plateforme-jee.html

Thunderbird & Lightning & Google Sync (compatibility 64 bit Linux)

If you run Thunderbird on a 64 bit Linux system you might have tried the calendar add-on "Lightning" and maybe even the synchronization add-on "Provider for Google calendar".
Linux (Ubuntu or Fedora) gets always the newest versions of Thunderbird via it's integrated package manager (Debian = apt-get; Red Hat = yum).
If you have had the problem that "Not compatible with Thunderbird x.x.x" please note that you can help yourself by downloading both add-ons from this special link:
1. Download specific Lightning
http://releases.mozilla.org/pub/mozilla.org/calendar/lightning/releases/1.0b2rc3/contrib/linux-x86_64/lightning.xpi
2. Download specific Provider for Google calendar

http://releases.mozilla.org/pub/mozilla.org/calendar/lightning/releases/1.0b2rc3/contrib/linux-x86_64/gdata-provider.xpi

1.0b2rc3 (change to latest appropriate version)

3. Install downloaded add-ons

Tools → Add-ons → Install → Choose the *.xpi → Restart Thunderbird

Other usefull links:

https://wiki.mozilla.org/Calendar:Lightning

https://wiki.mozilla.org/Calendar:GDATA_Provider

Adding CLASSPATH or JAVA_OPTS to JBoss

Summary:
This post shows howto add a new Java library (e.g. patched class) or a new Java startup option
to JBoss running under Linux


Main post:
Adding Java library
edit ./run.sh and add JBOSS_CLASSPATH="./lib/mypatchedJar.jar;$JBOSS_CLASSPATH"



Adding Java startup option

edit ./run.sh and add JAVA_OPTS="-D32 $JAVA_OPTS"

Howto install JBoss JON Server

Summary:
This is my quick and dirty installation guide for a JON Server and a JON Agent

Main post:
Set up database
- First you need to install a database (Oracle or PostGre are supported)

- yum install postgresql postgresql-server
(when using Fedora Linux, adapt for other distributions)

Configure the new service to start automatically
/sbin/chkconfig postgresql on

service postgresql initdb

change /var/lib/pgsql/data/postgresql.conf
listen_addresses='*'




http://www.postgresql.org/docs/9.0/static/auth-pg-hba-conf.html
/var/lib/pgsql/data/pg_hba.conf
host    all         all         0.0.0.0/0             trust

(by default it is only allowed to receive request from localhost:
host    all         all         127.0.0.1/32             trust)

service postgresql stop
service postgresql start


- Create a user: rhqadmin (password: rhqadmin)
http://docs.redhat.com/docs/en-US/JBoss_Operations_Network/2.4/html/Installation_Guide/JON_Server_Installation.html

cd /var/lib/pgsql/data
Create role rhqadmin: 

createuser -h 127.0.0.1 -p 5432 -U postgres -S -D -R rhqadmin

Create db rhq:

createdb -h 127.0.0.1 -p 5432 -U postgres -O rhqadmin rhq

(if you are using another machine to run your DB server make sure your SELinux is disabled)

service iptables stop
chkconfig iptables off


Installing JON server
Official Red Hat support

• Unzip the downloaded JON-zip-file
• $ cd /jon-server-2.x.x.GA/bin$ export JAVA_HOME=/usr/java/jdk1.6.0_05 $./rhq-server.sh start Trying to start the RHQ Server... RHQ Server (pid 14413) is starting
• Goto http://localhost:7080/ (and click "Click here to continue the installation")

Setting up JON database

• Click "Install"

        

           

          Screenshot of some tables created during installation

• You will be redirected to http://localhost:7080/Login.do

           

              Login with user rhqadmin (default password is rhqadmin)

              Itis not the user/pw of the database!

• A valid JBoss Operations Network license is required to use this software. Please click the Update License link below.

          

• Get your license here: https://access.redhat.com/jbossnetwork/restricted/softwareDownloadJBNLicense.html
• Click Browse ... and choose the downloaded xml license file. Click OK.

          

• If you want rather a command line installation than a GUI based installation, you could have entered all data within  /bin/rhq-server.properties (use generate-db-password.sh to generate an encrypted password). After that just run JON server : /bin/rhq-server.sh start

Installing the JBoss ON Agent
Official link

• Get the agent JAR by hitting http://localhost:7080/agentupdate/download
• Save the rhq-enterprise-agent-3.0.0.GA.jar (~ 11 MB) to the directory your agent should be installed to
• Run
  java -jar rhq-enterprise-agent-3.0.0.GA.jar --install
  which will create a rhq-agent directory
• In the directory ./rhq-agent/bin run
  ./rhq-agent.sh --cleanconfig

• After the first setup, start your agent with
  ./rhq-agent.sh

Further references:

More detailed installation steps you'll find here:

http://amitbhayani.blogspot.com/

http://pilhuhn.blogspot.com/2008/05/writing-rhq-plugin-part-1.html

http://community.jboss.org/wiki/JON23ControlGroupedServersViaTheCLI

The open source problem - "Unfakeable Linux"

Summary:
This post covers the problem of companies that provide open source software
Examples:
Red Hat Enterprise Linux --> CentOS
Red Hat Enterprise Linux --> Oracle Unbreakable Linux

Main post:
Open Source means anyone can take your source and redistribute it, as long as it is not sold.
Furthermore
CentOSOracle

What can comanpanies do?
1. Marketing
Red Hat "Unfakeable Linux" advertisement - smart answer to Oracle's copy of RHEL

Trying Java 7

Summary:
Being curious weather the new JConsole within JDK7 would still have problems connecting to JBoss JMX (check out older post), I gave a prerelease a shot and tried JDK7.


Main post:
Standard Edition 7 Binary Snapshot Releases

July 29, 2010
Build b103http://dlc.sun.com.edgesuite.net/jdk7/binaries/

Clustering JBoss: mod_jk VS mod_jk2 VS mod_proxy

Load balancing – JBoss Application Servers

Web Connectors	Apache version	JBoss platforms	Tomcat
mod_jk 1.2.x	Apache 2.	EAP4.3
mod_jk2	JBoss EAP 4.2	P2
mod_proxy	Apache 2.x (included in 2.2+)	EAP3.x
mod_cluster	Apache 2.2.10+	EAP 5.0.0+ & EWS 1.0+ EAP 3 + 4 (not certified)	Not certified

Parralel deployment in JBoss

Summary:
Howto hot-deploy a new version V2 of an application V1 without having a downtime of the application.

Main article:


References:
Weblogic

Separating log4j logging having the same WAR multiple times on one JBoss EAP 4.3

Summary:
This post will explain how to separate your log4j logs when having the same WARs deployed multiple times (packaged in an EAR) on the same JBoss.
For example myApp1.war and myApp1_copy.war.
This post is for JBoss EAP 4.3 ! For EAP 5.0.0 it does not work ! For EAP 5.1 you will not need this as it becomes simpler DeployURL equals the specific war name. (see end of post)

Main article:

While looking for a solution you will most likely find the wiki entry on jboss.org: http://community.jboss.org/wiki/SeparatingApplicationLogs

Unfortunately it does not cover the need to separate logging when having multiple equivalent WARs that log via the same class --> example1 (or that use a common .jar that does the logging --> example2)

example1.ear
├── myApp1.war
├── myApp1_copy.war
└── META-INF
    ├── application.xml
    └── jboss-app.xml

example2.ear
├── myApp1.war
├── myApp1_copy.war
├── loggingJar.jar
└── META-INF
    ├── application.xml
    └── jboss-app.xml

The log that you would want to see is:

JBOSS/server/<profile>/log/myApp1.log

JBOSS/server/<profile>/log/myApp1_copy.log

Adapting your log4J within JBoss

JBoss logging gives you the possibility to add filters within the jboss-log4j.xml file via the TCLFilter class. For the following DeployURL parameter it is usually sufficient to add a sub string of the deployed WAR name "myApp1-exp.war" (found in your JBOSS/server/<profile>/tmp/deploy directory). For our scenario this is not sufficient, we have to add /WEB-INF at the end of the string. If we do not add /WEB-INF we will get logs in both log files! The reason is that the TCLFilter class compares the value of DeployURL with the repositories of the Classloader ...
By the way if you want that JBoss takes changes of your jboss-log4j.xml into account immediately you'll want to set

The filter of your appender should look like this :

<filter class="org.jboss.logging.filter.TCLFilter">

    <param name="AcceptOnMatch" value="true"/>

    <param name="DeployURL" value="myApp1-exp.war/WEB-INF"/>

</filter>

To make sure that
myApp1.war logs into JBOSS/server/<profile>/log/myApp1.log
and 

myApp1_copy.log logs into JBOSS/server/<profile>/log/myApp1_copy.log

add the following into your jboss-log4j.xml:
<appender name="APPENDER-1" class="org.jboss.logging.appender.FileAppender">
        <errorHandler class="org.jboss.logging.util.OnlyOnceErrorHandler"/>
        <param name="File"        value="${jboss.server.home.dir}/log/web1.log"/>
        <param name="Append"      value="false"/>
        <param name="DatePattern" value="'.'yyyy-MM-dd"/>
        <layout class="org.apache.log4j.PatternLayout">
            <param name="ConversionPattern" value=" %r [%t] %-5p %c %x - %m%n"/>
        </layout>
        <filter class="org.jboss.logging.filter.TCLFilter">
            <param name="AcceptOnMatch" value="true"/>
            <param name="DeployURL" value="myApp1-exp.war/WEB-INF"/>
        </filter>
        <filter class="org.apache.log4j.varia.DenyAllFilter"></filter>        
   </appender>


<appender name="APPENDER-2" class="org.jboss.logging.appender.FileAppender">
        <errorHandler class="org.jboss.logging.util.OnlyOnceErrorHandler"/>
        <param name="File"        value="${jboss.server.home.dir}/log/web2.log"/>
        <param name="Append"      value="false"/>
        <param name="DatePattern" value="'.'yyyy-MM-dd"/>
        <layout class="org.apache.log4j.PatternLayout">
            <param name="ConversionPattern" value=" %r [%t] %-5p %c %x - %m%n"/>
            <!--param name="ConversionPattern" value="%d %-5p [%c] %m%n"/-->
        </layout>
        <filter class="org.jboss.logging.filter.TCLFilter">
            <param name="AcceptOnMatch" value="true"/>
            <param name="DeployURL" value="myApp1_copy-exp.war/WEB-INF"/>
          </filter>
        <filter class="org.apache.log4j.varia.DenyAllFilter"></filter>
</appender>

<category name="com.your.package"> 
            <priority value="DEBUG" /> 
            <appender-ref ref="APPENDER-1"/>
            <appender-ref ref="APPENDER-2"/>
</category>


The different JBoss EAP versions differ as the deployment process changes from each version:
EAP4.3:
<param name="DeployURL" value="myApp1-exp.war"/>
EAP5.0.0:
TCLFilter does not work at all here is the issue : https://issues.jboss.org/browse/JBAS-6532
EAP5.1: the TCLFilter can be set up with the name of the war
<param name="DeployURL" value="myApp1.war"/>

Network traffic tool Wireshark

Summary:
Installation problem of wireshark

Main post:
I tried to use wireshark the other day within a Fedora13_64 Gnome and ran into the following error:
# sudo yum install wireshark
Package wireshark-1.2.8-3.fc13.x86_64 already installed and latest version
# sudo wireshark
wireshark: command not found

The simple solution was, that it is not sufficient to have wireshark, you also need wireshark-gnome to get it working
# sudo yum install wireshark-gnome