Some useful unix commands i came across today

vmstat [pamam1] [param2]

shows the server statics along with swap space, cache,io and cpu utilization.

this stats are shown [param2] number of times with [param1] duration in between.

top

this command also shows the server statistics in detail with
contribution from each process.
This also shows the memory used,cpu utilization and command used for each thread.

finger

Shows the user who are currently logged into the node.

Simplest way to Configure Logging using Log4j

Ensure the log4j.jar & log4j.properties are in the classpath

Add the following into the required classes where logging is needed , so that log shows file name for each logger statement .

private static org.apache.log4j.Logger logger = org.apache.log4j.Logger.getLogger([classname].class);

folowed by logger.[info,debug,error,fatal etc]("logger message");

Contents in Log 4j.properties

log4j.rootLogger=ON, ROOT[enabling the root logger]
log4j.appender.ROOT=org.apache.log4j.DailyRollingFileAppender[saving logs on the daily basis with date appended]
log4j.appender.ROOT.File=/a/b/c.log[root log file path]
log4j.appender.ROOT.layout=org.apache.log4j.PatternLayout
log4j.appender.ROOT.layout.ConversionPattern=%d{dd MMM yyyy HH:mm:ss } %-5p [%t] %c{2} - %m%n [pattern for display in each line]

#FILE APPENDER [another logger file ]
log4j.appender.F2=org.apache.log4j.DailyRollingFileAppender
log4j.appender.F2.File=/a/b/d.log
log4j.appender.F2.layout=org.apache.log4j.PatternLayout
log4j.appender.F2.layout.ConversionPattern=%d{dd MMM yyyy HH:mm:ss } %-5p [%t] %c{2} - %m%n


log4j.logger.org.apache=ERROR[apache specific errors to be shown on both logger files]
log4j.logger.com.example=DEBUG,F2[for com.example modules debug status messages to be shown in second logger file]

For Log4j debugging
Append -Dlog4j.debug in JAVA_OPTS in catalina.sh which will enable detailed logs in catalina.out showing the logger configerations.

Apache Basics

Apache Restart

sudo /etc/rc.d/init.d/httpd restart

All requests with port :8080 are caught by tomcat as apache is listening at port 80

requests without ports are heard by apache which listens at port 80 and later redirects through 8009

Directory Structure

etc/httpd/

conf/ conf.d/ logs/ modules/ run/

conf/
httpd.conf ssl.crl ssl.crt ssl.csr ssl.key ssl.prm workers.properties

conf.d/

authz_ldap.conf jk.conf perl.conf php.conf python.conf README ssl.conf webalizer.conf welcome.conf


1. etc/httpd/conf/httpd.conf

Attributes include
DocumentRoot "/home/partyaccount/website/htdocs"
DirectoryIndex index.jsp index.html inde
DefaultType text/plain
ErrorLog logs/error_log
AddEncoding x-compress Z
ErrorDocument 500 "The server made a boo boo."

2. worker/properties

worker.list=worker1
worker.worker1.port=8009
worker.worker1.host=localhost
worker.worker1.type=ajp13
worker.worker1.lbfactor=1
[uri:/ws/*]
worker=ajp13:localhost:80


3. etc/httpd/conf.d/jk.conf

worker assighnments with respective .jsp,.html etc


4. etc/httpd/conf.d/ssl.conf

will have virtual hosts configured here ex:pa112 etc
document root
server admin
servername

Rewrite rules per vhost klike
RewriteEngine on
RewriteRule ^(.*)/.jsp$ $1.do [L]

Overwriting Struts Resouce Bundle Handling

Care needs to be taken when we overwrite the struts resource bundle handling,especially while overwriting the get text method in struts

This is because , if the customised gettext copies this statment

ResourceBundle bundle = LocalizedTextUtil.findResourceBundle(bundleName, resourceLocale);

where bundlename is the bundle in which property is expected

and if the property does not exist in that bundle, Struts internally looks for the entire set of bundles and searches the key. This is not what is expected
and it consumes a lot of time for each literal check and chokes the server.

A quick fix to this would be to search for the expected bundle and then check if the key which is expected is present.This can be done by using bundle.containskey() and later if it exists then by using the above statement. This reduces a lot of time wasted checking for the entire set of bundle for each property.

Your Kit : A CPU and memory Java Profiler

This is a simple kit with 15 day trial period licence which can be got from http://www.yourkit.com/.

It has both windows and linux versions.For the linux version just copy the

1. copy the libyjpagent.so into tomcat/bin

2. set the path by typing export LD_LIBRARY_PATH=/tomcat/bin

3. Add -agentlib:yjpagent to Tomcat JVM startup params as

JAVA_OPTS=" -agentlib:yjpagent -server -Xms512m -Xmx768m ......" in catalina.sh


To connect for the same from Windows

1 Install the windos version

2 Click on connect remotely . Enter the ip of the profiling machine

3 This gives most of the stats. For time wise distibution of thread on the running methods

click start profiling.

It also gives snapshots for the required time duration.

Adding Logger interceptor in Struts

When i had a requirement to add user name and session id to every logger statement for debugging purposes i did not think it would be so easy. I came across sites which suggested adding some extra modules and configurations into the log4j.

But i realized there was a much simpler way which was to modify the thread name appending application specific user name and session id into it. As anyways log4j will print the thread name at every logger statement . Thus with minimal changes my cause was served. I added this functionality into every request by making it as a interceptor and adding it into the basic stack.
I have attached the interceptor code for reference.

Tasks get easy when seen for different perspectives.

Sample WorkFlow Engine Using Enhydra Shark - Struts 2

You can download the sample code(Unzip the war) from

http://www.esnips.com/doc/2d2cc8a2-d336-454b-bb76-5950a22714b9/Enyhydra