This blog is all about me doing nothing serious :)

Creating large dummy files for test

2013-03-06T14:01:00.000+01:00

At times during development we need to create some large files regardless of what data is there inside. For e.g. to see if your program can handle insanely large file or to see how long it takes to transfer a particular size on a network or on your JMS queue etc. For such matter almost all OS provides some nice ways - which sadly I was not aware till now :-(. Anyway here are the different ways you can create such files under different OS

Windows:

Open the command shell - cmd
Execute the following command in order to create a file of 1mb
fsutil file createnew ./test.tmp 1048576

Linux:

User convert and copy command - dd
dd if=/dev/zero of=test.tmp bs=1M count=1

OS X & Solaris:

mkfile 1m test.tmp

Go fill your disk with insanely large files :-)

Cheers !!!
- Jay

Import Oracle dump with different table space, schema , roles with imp utility

2013-02-08T12:30:00.001+01:00

At (bad)times it happens that for a nasty bug happening only in client environment, we receive dump file (exported using Oracle's exp utility) from client or say from different environment where table space, schema and roles are different from the one we use in development environment. In such a case it becomes difficult or impossible to import such dump file using oracle's imp utility. To tackle such situation there is a kind of workaround.

Note: Base assumption over here is you already have a local database in your development environment having the structurally same schema as you wish to import from the dump file. If this is the case below are the steps

Disable all the constraints: Below script will give you ALTER TABLE scripts to disable all the constraints in your schema (the one to which you want to import the data from dump)
SELECT 'ALTER TABLE'||' '||owner||'.'||table_name||' DISABLE CONSTRAINT '||constraint_name||' ;' FROM dba_constraints WHERE constraint_type = 'R' and owner='<OWNER>' and status = 'ENABLED';
Truncate all the tables: Below script will give you TRUNCATE TABLE scripts for all the tables in your schema (the one to which you want to import the data from dump)
SELECT 'TRUNCATE TABLE '||OWNER||'.'||TABLE_NAME||' ;' FROM DBA_TABLES WHERE OWNER='<OWNER>';
Import data from the dump: Now you have a clean schema with only structure. Now you need to import ONLY data from the dump. No structure, constraints etc. Below is the script to import only data using imp utility.
imp <USER>/<PASSWORD>@<SCHEMA_NAME> file=<DUMPFILE_NAME> INDEXES=y full=y data_only=y
Enable all constraints: Below script will give you ALTER TABLE scripts to enable all the constraints in your schema
SELECT 'ALTER TABLE'||' '||owner||'.'||table_name||' ENABLE CONSTRAINT '||constraint_name||' ;' FROM dba_constraints WHERE constraint_type = 'R' and owner='<OWNER>' and status = 'ENABLED';
Done

Cheers !!!
- Jay

Ref: Got scripts in step 1, 2 and 4 from Sergey. Thanks Sergey

Importing Eclipse Workspace Settings From An Old Workspace

2012-10-01T16:23:00.000+02:00

This is a classic problem I guess any java developer or better say any eclipse developer would have faced or is facing every other day. I've been using eclipse as my primary IDE for years and like most of the users, I also have multiple workspaces to manage my code base depending upon the trunk/branch or any particular release I'm working on. Having multiple workspaces makes life so easy and work so hassle free, but it also comes with some "extra overhead" of having consistence settings across the workspace and at time across the instances (home/work). Until sometime back I used to set all the configuration (except formatters) each time I create new workspace and also had to update any settings I change in workspace to another. Recently I found a easy way to have same settings in matter of seconds from an existing workspace to the new workspace. This is what I'm going to show you here in this post. To do so follow the below mentioned simple steps.

Create the new workspace.
Close your eclipse instance having this new workspace open.
Go to your old workspace file system from which you want your preferences Copy the directory <oldworkspaceroot>\.metadata\.plugins\org.eclipse.core.runtime\.settings to your <newworkspaceroot>\.metadata\.plugins\org.eclipse.core.runtime\
Start your eclipse.
Done!!!

Cheers !!!
- Jay

Error: could not open `C:\Program Files\Java\jre6\lib\amd64\jvm.cfg'

2012-09-25T11:55:00.000+02:00

In some cases where you are playing around with your JDK/JRE its possible you corrupt your settings or JRE in total. I personally dont like the default JRE installation location on windows i.e. C:\Program Files\ for the fact that there is a space between Program and Files which at times create problems especially when you are using cygwin and writing some linux script. So as a regular practice I always install my JDK/JRE to c:\java. So the other day I got this brand new Windows 7 64bit machine with 8gb RAM and 3.3ghz i5 processor. My system admin was so kind enough that he gave me all the *required* tools already installed for my development. This obviously came with JRE 6 installed in Program Files directory. So I just un-installed it, got new 64bit JDK and JRE from Oracle and installed it under my regular directory i.e. c:\java. Now when I turned to my cygwin terminal and ran java what I see is below mentioned error

Error: could not open `C:\Program Files\Java\jre6\lib\amd64\jvm.cfg'

I checked my environment variables - JAVA_HOME & PATH and they all refer to c:\java. So this was bit frustrating. After sometime I found that the default installation also copied java.exe, javaw.exe and javaws.exe to C:\Windows\System32 (i.e. uninstall of JRE didnt went well). I just removed them and voila, I'm back on track. That annoying error is no longer popping.

Cheers !!!
- Jay

WordPress on OS X : Error establishing connection mysql

2011-12-04T19:38:00.001+01:00

Recently I was trying to install/configure WordPress on OS X (Lion) with MySql 5.5. After reading the installation guide, I did not expect anything to go wrong, but as it goes without saying, theory is totally different from practical, I did face issue. After setting wp-config.php, when I hit the /wordpress/wp-admin/install.php, I got a nice page saying Error establishing a database connection. Obvious reason for this would be (as mentioned on the error page) would be the wrong user/password or wrong host or wrong database name. I checked all parameters and none were wrong and I was also able to connect to MySQL database which I created for WordPress with MySQL WorkBench and also from mysql command line.

To verify my settings, I just created and a test php page to test if I can connect to my newly create WordPress database. I was not able to connect and got some more details about no connection. I got following wraning

ErrorException [ Warning ]: mysql_connect() [function.mysql-connect]: [2002] No such file or directory (trying to connect via unix:///var/mysql/mysql.sock)

Obviously when I checked /var/mysql/mysql.sock, I could not find it. So next step was to find where is my missing mysql.sock and found it under /tmp. After googling a lot, I could not find any concrete solution to resolve this. Niether there is anything which I can change for MySQL configuration nor for the PHP configuration. So my last resort and the easy solution was creating a symbolic link. I just created a symbolic link on /var/mysql to point to /tmp (sudo ln -s /tmp /var/mysql) and hit the refresh on WordPress installation page and everything was just smooth. Happy blogging :-)

Cheers !!!
- Jay

Android Emulator Error unknown virtual device name

2011-05-27T15:44:00.000+02:00

Couple of days back I got a windows system where I was required to do some changes for a Android app. So I started by downloading Android SDK and required packages. I created a AVD using SDK Manager. When I tried to start the AVD I got an error emulator: ERROR: unknown virtual device name. Below is the snapshot for the same.

At that time I googled a bit and found the solution by reading couple of articles. Today again one of my colleagues faced the same issue and I was scratching my head to find out what was the solution. Again wasted around 30mins to find the solution. Finally it got working, but 30mins gone wasted on a problem for which I already searched and found solution in past. So just to avoid any such instance, as a note-to-myself, in this article I'll document the exact steps required to fix this issue

Reason for this issue is Android SDK Manager (by default) expects the AVDs to be in C:\Documents and Settings\<win_user>\.android (in case of win xp) directory when AVD is started. But when AVDs are created (by default) SDK Manager creates it under C:\WINNT\.android directory. Here <win_user> is the name of the user using which you have logged in to Windows PC.

To resolve this, one of the solutions is to create a symbolic link to C:\WINNT\.android directory. To do so, follow below mentioned steps

Download Sysinternal's Junction utility.
Create a symbolic link using below command
junction "C:\Documents and Settings\<win_user>\.android" C:\WINNT\.android
Try starting your AVD now. It should start
Done

Cheers !!!
- Jay

Installing Evolus Pencil Extention On Firefox 4

2011-04-05T22:56:00.002+02:00

Recently we got a fabulous update of Firefox 4. With many good features and nice UI, we also get to see many extensions stop working. One of them is Evolus Pencil Sketch. Why, because the max version defined in metadata is 3.6.* and that stops the extension from installation.

To get this thing working...

Download the Pencil-1.2-0-fx.xpi
Unzip it with your favorite zip archiver
Open install.rdf
Locate the string - em:maxVersion="3.6.*"
Replace it with - em:maxVersion="4.*"
Save the file. Now your file should look something as shown in below screenshot
Now archive all the files - install.rdf, update.rdf & chrome.manifest - and the directory - chrome - into one zip file and name it as Pencil-1.2-0-fx.xpi. It does not necessarily has to be Pencil-1.2-0-fx.xpi, you can have it as MyFile.zip or anything of your choice.
Now open up Firefox 4, go to Tools menu and select Add-ons
Click Install Add-on From File...
Select the archive you just created and you are done

One Click Lock For Mac OSX

2011-03-14T17:29:00.000+01:00

Since long, I was looking for some way to lock my MBP with keyboard shortcut (similar to windows - WIN + L). I googled a lot and found many suggestions from many users and also some utilities too :-). Finally now I got a clean solution to do this. Thanks to Karel who wrote a small how-to on this. If you are interested, have a look at his blogaria http://kubat.nl/pages/blogaria/239#239.

Cheers !!!
- Jay

MySQL 5.5 Startup Problems On OS X 10.6.5

2010-12-25T15:38:00.003+01:00

Today morning I downloaded MySQL Community Server 5.5.8 for OS X 10.6.5. I downloaded 64bit disk image and started installation by first installing mysql-5.5.8-osx10.6-x86_64.pkg, followed by startup item - MySQLStartupItem.pkg - and then the system pref pane - MySQL.prefPane. Everything went well with the installation. After that I opened System Preference and clicked on MySQL pane to start the server. When I clicked on Start MySQL Server button, nothing happened. Neither server started nor did I get any confirmation message or any error message. Reason for such failure to start is the server startup script - mysql.server - is missing the MySQL installation directory path as well as MySQL data directory path. To resolve this locate basedir= and put entry as shown below and save the file. mysql.server file can be found under /usr/local/mysql/support-files directory

basedir=/usr/local/mysql
datadir=/usr/local/mysql/data

Alternatively above entries can also be put in /etc/my.cnf file too. Better place to put this instead of modifying packaged/default startup script - /usr/local/mysql/support-files/mysql.server. Anyway its individual choice.

Now try to start the server by clicking Start MySQL Server button in MySQL System Preference pane. I could start the server and can create a new server instance using MySQL Workbench. I was a happy man. But my happiness did not end long. As soon as I restarted my laptop, I got an annoying alert...

Insecure Startup Item disabled. “/Library/StartupItems/MySQLCOM” has not been started because it does not have the proper security settings.

This was because an entry in /etc/hostconfig as MYSQLCOM=-NO-. Due to ownership issue on startup item MySQLCOM, above alert was flashing on OS startup. To fix this just change the ownership of MySQLCOM startup item as shown below

sudo chown -R root:wheel /Library/StartupItems/MySQLCOM

To test just logoff or restart the system and that Insecure Startup Item disabled alert should not come again.

Cheers !!!
- Jay

Adding spring-ws-archetype In STS 2.5.1 / Eclipse

2010-12-20T22:18:00.002+01:00

I recently started working on Spring-WS. I'm already using Spring 3.0, so as IDE for development my obvious choice was STS. I downloaded the latest version 2.5.1 and launched the IDE. As mentioned on Spring reference documentation, I started with creating a new maven project in STS. Soon I realized that STS (at least version 2.5.1) doesn't come bundled with spring-ws archetype. STS requires this archetype to be added manually. It took me while to recall how to add a new archetype to STS (or say Eclipse). After a couple of minutes of head scratching I could recall. So for my future reference and probably for others help here is how you add a new archetype in STS OR Eclipse

Step 1: Open STS
Step 2: Start New Project wizard and select Maven Project
Step 3: Click Next on Select project name and location dialog
Step 4: On Select an Archetype dialog click Add Archetype... button. This will open Add Archetype dialog

Select an Archetype Dialog

Step 5: In Add Archetype dialog add Archetype Group Id, Artifact Id and version as shown in below snapshot and click OK button

Add Archetype Dialog

Step 6: On click OK button STS will download required files from server and new archetype - spring-ws-archetype - will be added to already bundled Nexus Indexer Archetype catalog as shown in below snapshot

Added spring-ws-archetype

Step 7: Start creating project with newly added archetype.

Step 8: Done !!!

Note: MVN Repository - http://mvnrepository.com/ is a very good reference to find details of any maven artifacts and versions. I used the same to add details for spring-ws-archetype.

Cheers !!!

- Jay

Upgrading Ubuntu To New Release Behind Proxy

2010-11-13T23:44:00.001+01:00

At my work place I connect to internet via proxy (requiring username/password). I've one desktop running Ubuntu 10.04 and wanted to upgrade to 10.10 the latest long term release. Even after applying all the updates shown in Update Manager, I never get to see the availability of new release.

After lot of googling and playing around with proxy settings in Synaptic Package manager and also in Network Proxy, I found no success. On various forums also I see many facing the same problem. After some days of such struggle and trying different options to make update manager "read" my proxy settings I realized that Update Manager correctly detects the proxy settings to show the available updates, but the apt-get which actually checks for availability of new release does not "read" the proxy settings from the same location as Update Manager does.

So as suggested on Ubuntu help pages on setting up proxy, I tried to temporarily setup proxy by exporting http_proxy and then tried sudo apt-get update. Still the result was same. Still no option to upgrade to 10.10 :-(

After this I tried to change the proxy settings in /etc/apt/apt.conf to provide proxy along with the username/password. After saving the changes, apt-get successfully updated all the packages from the repositories and Update Manager showed me the 10.10 availability message.

So even though Ubuntu help page says that you can set the proxy temporarily by exporting http_proxy, it seems apt-get ignores http_proxy and always refers to atp.conf (/etc/apt/apt.conf). Well things are not always as it should be and it applies to worlds one of the best OS too.

Cheers !!!
- Jay

Using MyBatis 3 with Spring 3 Using Mapper XML Files

2010-11-13T22:34:00.005+01:00

Recently I started to work on design an application as a part of a new project assigned to me at my work place. Since this is totally a new application to be developed from scratch, I got liberty/flexibility to choose the tools/frameworks which will form the foundation of the application. With lots of frameworks to choose from and many choices available for different tiers, I started evaluating each framework against the requirement of the application vs where it fits in which tier. When I came to data access tier (resource tier as some say), I realized that MyBatis (new name of iBatis after it moved to Google code from Apache) was best fit as ORM framework. I also decided to use Spring 3.x as DI framework and we all know how well MyBatis/iBatis and Spring (beans, data access, transaction etc) gel with each other. Using Spring's DI, declarative transaction management and MyBatis' query externalization, we as a developer can concentrate on implementing application/business logic and keep rest of the stuff like transaction management, JDBC resource management etc to be "headache" of container/frameworks.

So as I decided to go with MyBatis, I came to know that Spring 3.0 development finished first then MyBatis 3 and hence there is no official support in Spring 3.0 for MyBatis 3.0. Its expected to be there in Spring 3.1 release (Source: Spring-MyBatis integration RC2 reference doc). But thanks to MyBatis community, there is already a project for myabtis-spring integrations. At start I struggled a bit on how to wire Spring beans, dao with MyBatis mapper configuration files. As I was googling around, I stumbled upon a very good post by Giovanni Cuccu on this. It was very enlightening and provided me a good start. I created a maven/spring/mybatis project in my eclipse and started writing down some quick and crude code to test out. As soon as I tried to run my sample code I encountered...

java.lang.IllegalArgumentException: Property 'sqlSessionTemplate' is required

After debugging the mybatis-spring integration source code, I found that since I was using Spring DAO using mybatis-spring integration, org.mybatis.spring.support.SqlSessionDaoSupport class' checkDaoConfig explicitly checks for the sqlSessionTemplate property to be defined. After trying for some time and to no success, I changed my DAO implementation to be a mapper interface using MapperFactoryBean. This eliminated of having a class implementing SqlSessionDaoSupport just to invoke Spring's SqlMapClientTemplate's helper methods. Instead now its just an interface holding method signatures mirroring the java representation of mapped sql statements in MyBatis mapper configuration files. With this changed implementation, everything worked like a charm.

Here is the spring wiring I did to make mybatis-spring work with mapper files. Usually I keep my spring configuration logically separated by keep configuration related to data access in one file and application DAOs go to another. So with this I end up creating four configuration files and one java interface for this project.

data-access-spring-context.xml: Holding spring configuration defining my data access
mybatis-config.xml: Standard MyBatis configuration file
user-dao.xml: MyBatis mapper file. One mapper file for one DAO defined in dao-spring-context.xml
dao-spring-context.xml: Holding all DAOs used by application
UserDAO.java

Version of various plugin and framework I used for this little POC...

Spring 3.0.5
MyBatis 3.0.2
MyBatis-Spring 1.0.0-RC2

Let's start visiting each of these files to understand the required configuration...

data-access-spring-context.xml:



    
        
            classpath:jdbc.properties
        
    

    
        
            ${DataSource.JndiName.DS}
        
        
            false

For simplicity, I've not included the transaction manager, advice point cut etc in this. So here, main point of interest is how the sql session factory and sql session template is configured which actually wires MyBatis configuration to Spring data access

mybatis-config.xml:
```
    
        
    
    
        
    
```
This is a standard MyBatis configuration file, holding all the mappers of the applications.
user-dao.xml
```
    
        
        
    

    
```
Here I've defined a SQL queries required my my User DAO. User DAO is wired to spring data access and mybatis in my DAO spring context i.e. dao-spring-context.xml
dao-spring-context.xml:
```
    
        
        
    
```
Notice that the fully qualified interface name defined as mapperInterface serves as the namespace for the mapper in user-dao.xml.

UserDAO.java:

package com.jak.sandbox.biz.dao;

import java.util.List;

import com.jak.sandbox.shared.dto.model.User;

public interface UserDAO
{
    List<User> selectUsers();
}

Well this is pretty much you need to do in order to use MyBatis with Spring. Now just write a test program which loads both spring context as defined above, inject userDAO bean to any of your Java class (typically a business service) and invoke selectUsers method.

MyBatis & Spring made data access so simple and using interface as your data access and having implementation as MyBatis mappers, makes data access more cleaner. This is because ideally your data access should only be taking request from business tier to perform some operations like select/insert/update/delete and return the result of such operation back. There should not be any logic. Having your data access as interfaces, takes the any chances of developer mistakenly putting any logic.

Cheers !!!
- Jay

With JavaScript Android Emulator Webkit Is More Forgiving Then Webkit on Phone

2010-11-10T00:21:00.025+01:00

Recently I was developing a mobile application on Android platform which involves various widgets of Android along with some stuff with JavaScript and WebView (Would like to write a post on my experience with Android development - may be later). Actually I wanted to plot some graphs and for that matter I opted for jQuery and flot plugin. Flot is an amazing plugin you can get to plot some really professional graphs. Well, while plotting some of the graphs, I started with some JavaScript code and writing a JavaScript handle which basically just creates JSON objects and arrays to use as my input data to flot to plot the graphs.

As most of the Android developers, I also use Android emulator to test my code. Once I have a bit of working code, I put it on my HTC Desire to get the actual feeling. So with these graph, I wrote couple of JS function along with JS handler, I deployed my first cut of the graph to my Android 1.6 emulator (API Level 4) and launched the application. Graph got plotted as expected. So far so good. I added couple of improvements, like enabling hovers, clicks, legends etc and then again tested on emulator. It gave expected results. So now it was time to deploy it on my HTC Desire. After deploying it, when I launched the application, I could just see the white screen of WebView and nothing else. No beautiful graphs. Checked LogCat, Eclipse console, everything was fine. No error/warning messages. I again tested the same on emulator and it was working like a charm.

After doing a lot of debugging with JavaScript code (let me confess, it was more of hit-and-try) I found that I declared couple of JS variables like

    var x = 0,
    var y = 1

Notice the comma after var x = 0 ? Ideally it should be semicolon but as all developers do silly mistakes, I also had slip of finger. This was not highlighted by the Eclipse HTML editor and it worked quite well on emulator. But when it came to actual phone, this was caught and it stopped everything. Liked that WebKit/WebView is very strict, but at the same time emulator behaved differently. Also there were no logs logged in LogCat. This behavior of emulator really scared me. I know it was a mistake on my part, but at least emulator should behave in same manner as actual software does. I sincerely hope that with new upcoming updates (may be with Gingerbread) this kind of issues are resolved and we can see the consistent behavior of emulator and phone software.

Cheers !!!
- Jay

Problems With Ubuntu 10.x USB Installation

2010-10-30T09:49:00.003+02:00

Recently I was trying to install Ubuntu 10.10 on one of my friends' Dell mini notebook after his windows 7 starter crashed horribly and left him with black/blue screen and no more boots. After trying for 2-3 days to get windows back to life without any success, my friend heard some golden worlds to move to Ubuntu. We decided to get his mini a great OS. Since I ran out of blank CD/DVD, we decided to create a bootable USB with Ubuntu installer on it and install Ubuntu over the weekend.

So it was Friday night where we were quite excited to get Maverick Meerkat installed and to start with downloaded the ISO image of it from Ubuntu site. I never tried a USB install so was bit skeptic too on if it will go well on my friend's Dell mini. For this my dear friend also bought a brand new 4GB Duracell USB drive on which we decided to make a startup disk. We started following the guidelines as mentioned on Ubuntu download page. After download was complete, my friend's brand new USB drive was deflowered by creating the Ubuntu startup disk. So far so good, all went well.

After ensuring that USB device is the first in boot sequence, we started the notebook. Within seconds we saw Ubuntu up and running from USB with an icon to start installation on the desktop. Without further ado, we started the installation by clicking the install icon. It first asked for the language to select, then timezone and third was keyboard selection. So far so good, everything was going quite well. After we hit forward button from Keyboard selection, nothing happened except the tiny loading image circulating.

After sometime when nothing happened, I turned to my dear friend - Google - for rescue. Tried couple of "smart solutions" posted by many and to no help, I found a post with an unusual solution. Change the USB drive you are using as startup disk. Can you believe this? But as you know a frustrated person will do anything to get things work, I took out my PNY Micro Attaché, dumped all crap from it to my laptop and created a fresh startup disk. With very less hopes we booted friend's notebook with it and restarted the installation. To my surprise installation progressed like a charm. I did see the partition manager after clicking forward on keyboard selection dialog and rest of the installation went well.

At the end, this whole process brought my friend again back to life when he could surf and do all stuff on his notebook again and left me with many questions to which I'm still trying to understand such a behavior. What possibly could go wrong with installation and what it has to do with the make of a USB drive. If system can be booted from a USB drive and Ubuntu just working fine booted from USB then why installation can not go ahead with the same USB drive? Let me know if you have any clue or a proper answer to this.

Cheers !!!
- Jay

Struts 2.x Application Deployment On WAS 6.1

2010-07-16T17:59:00.004+02:00

Recently one my colleagues - @chaituvj - was trying to deploy a Struts 2.1 web application on WAS 6.1. Upon hitting the first action (*.action), page showed HTTP 404: File Not Found error and WAS threw following exception

java.io.FileNotFoundException: /login.action
      at com.ibm.ws.webcontainer.extension.DefaultExtensionProcessor.handleRequest(DefaultExtensionProcessor.java:414)
      at com.ibm.ws.wswebcontainer.extension.DefaultExtensionProcessor.handleRequest(DefaultExtensionProcessor.java:113)
      at com.ibm.ws.webcontainer.webapp.WebApp.handleRequest(WebApp.java:3276)
      at com.ibm.ws.webcontainer.webapp.WebGroup.handleRequest(WebGroup.java:267)
      at com.ibm.ws.webcontainer.WebContainer.handleRequest(WebContainer.java:811)
      at com.ibm.ws.wswebcontainer.WebContainer.handleRequest(WebContainer.java:1455)
      at com.ibm.ws.webcontainer.channel.WCChannelLink.ready(WCChannelLink.java:113)
      ...

Same war file works like a charm on JBoss and Tomcat container, but when it comes to WAS 6.1, it just cant find the struts action URLs.

Upon reviewing application code, configurations a zillion time we turned to the life saver - Google. Instantly we realized that we are not the only unlucky soles. Upon going through many posts we found that WAS 6.1 web container by default does not invoke any of the custom servlet filters defined in web.xml. This means that it does not invoke the Struts 2 filter and hence Struts application context was never initialized (Smart IBM guys). To change this behavior and enable invocation of custom filters, a custom property needs to be defined for the web container under WAS 6.1. Here is what you need to do...

Open the WAS admin console.
Select the application server where you are deploying the *ill-fated* Struts 2 application
Select Web Container Settings
Select Web Container
Select Custom Properties
Define new property com.ibm.ws.webcontainer.invokefilterscompatibility and set its value as true
Restart application server
Done !!!

With above change your Struts 2 web app should start working with WAS 6.1

Cheers !!!
- Jay

Connecting MySQL 5.1 Database Remotely Using JDBC

2010-06-22T17:58:00.017+02:00

Recently I installed MySQL 5.1 on my Ubuntu Lucid Lynx and then was trying to connect to database from my windows machine and stuck with following exception

com.mysql.jdbc.CommunicationsException: Communications link failure

Last packet sent to the server was 0 ms ago.
     at com.mysql.jdbc.SQLError.createCommunicationsException(SQLError.java:1070)
     at com.mysql.jdbc.ConnectionImpl.createNewIO(ConnectionImpl.java:2104)
     at com.mysql.jdbc.ConnectionImpl.(ConnectionImpl.java:729)
     at com.mysql.jdbc.ConnectionImpl.getInstance(ConnectionImpl.java:298)
     at com.mysql.jdbc.NonRegisteringDriver.connect(NonRegisteringDriver.java:283)
     at java.sql.DriverManager.getConnection(DriverManager.java:512)
     at java.sql.DriverManager.getConnection(DriverManager.java:171)
     at com.jak.sandbox.hsm.HSMTest.testJDBC(HSMTest.java:35)
     at com.jak.sandbox.hsm.HSMTest.main(HSMTest.java:24)
Caused by: java.net.ConnectException: Connection refused: connect
     at java.net.PlainSocketImpl.socketConnect(Native Method)
     at java.net.PlainSocketImpl.doConnect(PlainSocketImpl.java:305)
     at java.net.PlainSocketImpl.connectToAddress(PlainSocketImpl.java:171)
     at java.net.PlainSocketImpl.connect(PlainSocketImpl.java:158)
     at java.net.Socket.connect(Socket.java:452)
     at java.net.Socket.connect(Socket.java:402)
     at java.net.Socket.(Socket.java:309)
     at java.net.Socket.(Socket.java:153)
     at com.mysql.jdbc.StandardSocketFactory.connect(StandardSocketFactory.java:256)
     at com.mysql.jdbc.MysqlIO.(MysqlIO.java:276)
     at com.mysql.jdbc.ConnectionImpl.createNewIO(ConnectionImpl.java:2027)
 ... 7 more

After googling a lot and reading may forums finally found solution from Ubuntu Forums. Here is the deal

By default MySQL 5.1 allows only local connection
Remote connections should explicitly be turned on

As with the previous version of MySQL, where server starts with --skip-networking to prevent all TCP/IP connections, MySQL 5.1 does this by means of configuration in my.cnf file. This file stores all the configuration on how server starts. Default configuration is to prevent all TCP/IP connections. This means that if you want your JAVA application to talk to MySQL database, you will see the above mentioned exception. This default configuration is displayed in below snapshot...

So to allow TCP/IP connections just comment out the line bind-address = 127.0.0.1 and restart the server. Settings mentioned here opens up connection from any IP. So not might be a very good configuration if we think in terms of database administration or security. But for a developer like me, this suffices the requirement. Regarding database administration and security, let me not poke my nose and leave it upto smart database administrators.

[UPDATE - 22 July 2010 16:48] - Found same issue (rather default configuration) for JBoss too. When you run JBoss application server with default configuration, it will not be accessible over internet/intranet. It will be accessible only on localhost. To make it available over the internet/intranet, start JBoss with run -b 0.0.0.0 and it should work.

Cheers !!!
- Jay

Setting Reverse Proxy With Apache2 on Ubuntu (Debian)

2010-04-09T14:52:00.031+02:00

Apache HTTP server is probably the most used open source free web server and does not require any introduction. Also setting up reverse proxy using Apache is also not new. In fact this blog entry does not talk about anything new. Over here I'm just jotting down easy to follow steps to setup a reverse proxy using Apache 2.2 on Ubuntu or any Debian flavor.

Main reason for writing this is, I've basically used Apache2 on windows and configuration is bit different then how it's on Debian. In windows typically you will find all configuration under httpd.conf file where you can load any particular module using LoadModule. This is not how Apache2 works exactly with Debian. With a typical install, you will find httpd.conf totally empty. Also loading modules dynamically does not work by having LoadModule declaration.

Without going much into details of these differences, I'll just write down how to setup reverse proxy...

Load required modules - proxy, proxy_http, proxy_connect

sudo a2enmode proxy_connect
sudo a2enmode proxy_http
Above will automatically enable proxy module.

Configure Site - Change your directory to the site for which you want to setup reverser proxy. For simplicity, I'll refer to the default site located at /etc/apache2/sites-enabled. You need to have root privilege to modify the site configuration. Below is the typical content for the default site i.e. 000-default

<virtualhost *:80>
    ServerAdmin webmaster@localhost

    DocumentRoot /var/www
    <directory />
        Options FollowSymLinks
        AllowOverride None
    </Directory>
    <directory /var/www/>
        Options Indexes FollowSymLinks MultiViews
        AllowOverride None
        Order allow,deny
        Allow from all
    </Directory>

    ScriptAlias /cgi-bin/ /usr/lib/cgi-bin/
    <directory "/usr/lib/cgi-bin">
        AllowOverride None
        Options +ExecCGI -MultiViews +SymLinksIfOwnerMatch
        Order allow,deny
        Allow from all
    </Directory>

    ErrorLog /var/log/apache2/error.log

    # Possible values include: debug, info, notice, warn, error, crit,
    # alert, emerg.
    LogLevel warn

    CustomLog /var/log/apache2/access.log combined

    Alias /doc/ "/usr/share/doc/"
    <directory "/usr/share/doc/">
        Options Indexes MultiViews FollowSymLinks
        AllowOverride None
        Order deny,allow
        Deny from all
        Allow from 127.0.0.0/255.0.0.0 ::1/128
    </Directory>

</VirtualHost>

To setup a reverse proxy for URL http://localhost:8080/myapp/ please add following lines at the end of the VirtualHost tag as displayed above...

ProxyPass /myapp/ http://localhost:8080/myapp/
ProxyPassReverse /myapp/ http://localhost:8080/myapp/

Enable access - To allow access for all requests to this proxy, edit /etc/apache2/mods-available/proxy.conf. By default access to all is denied. To enable the access, put line Allow from all. Typical proxy.conf after this change will look something like...

<IfModule mod_proxy.c>
    #turning ProxyRequests on and allowing proxying from all may allow
        #spammers to use your proxy to send email.

        ProxyRequests Off

        <Proxy *>
                AddDefaultCharset off
                Order deny,allow
                Deny from all
                Allow from all
        </Proxy>

        # Enable/disable the handling of HTTP/1.1 "Via:" headers.
        # ("Full" adds the server version; "Block" removes all outgoing Via: headers)
        # Set to one of: Off | On | Full | Block

        ProxyVia On
</IfModule>

Note the line ProxyRequests Off. This is important in order to enable reverse proxy

Restart Apache2 - To get all of above configuration changes in effect, have a clean restart. Recommended way is...

sudo apache2ctl graceful
Done !!! Happy reverse proxying

Cheers !!!
- Jay

Using SmartVoip on Ubuntu 9.10 using Linphone

2010-03-23T23:40:00.008+01:00

Since long time I was looking a way to use SmartVoip on Ubuntu. I searched some forums just to discover that there are many who are looking for the same. One of the articles mentioned that Ekiga Softphone shipped with default Ubuntu installation can be configured to use SmartVoip SIP details, but unfortunately I could not manage to configure it 100%.

Finally today I managed to find how using Linphone I can use SmartVoip SIP details and make calls. In this article I'll mention how to configure Linphone to use your SmartVoip account details on Ubuntu 9.10 a.k.a "Karmic Koala" with a hope that it will help some poor soul like me who is looking for really easy to configure softphone to use SmartVoip on Ubuntu. So here it goes...

Get Linphone. Either download it from http://www.linphone.org/ OR use Ubuntu Software Center to download and install it easily.
Once installed open Linphone...
Open preferences from Linphone menu, select "Manage SIP Accounts" and click Add button
In above "Configure a SIP account" enter information of your SmartVoip account as below...

Your SIP identity: sip:<SmartVoip UserName>@smartvoip.com. Replace <SmartVoip UserName> with your SmartVoip user name.
SIP Proxy address: sip:sip.smartvoip.com
Leave rest of the fields as is and click Ok.
Your SmartVoip SIP details are now configured with Linphone.

Now you can start making calls to other SIP phone or call to any mobile or land line number using your SmartVoip account.

Cheers !!!

- Jay

Life Cycle Is All Backwards

2010-02-06T16:33:00.008+01:00

Have you ever wondered what would be like if we were born old and die as a baby? George Costanza describes it best. Enjoy it...

The most unfair thing about life is the way it ends. I mean, life is tough. It takes up a lot of your time. What do you get at the end of it? Death! What’s that, a bonus?! I think the life cycle is all backwards. You should die first, get it out of the way. Then you go live in an old age home. You get kicked out for being too healthy, go collect your pension, then, when you start work, you get a gold watch on your first day. You work forty years until you’re young enough to enjoy your retirement. You drink alcohol, you party, and you get ready for high school. You go to primary school, you become a kid, you play, you have no responsibilities, you become a little baby, you go back, you spend your last 9 months floating with luxuries like central heating, spa, room service on tap, then you finish off as an orgasm! :-)

Cheers !!!
- Jay

Searching classes in a JAR or multiple JARs

2010-01-24T09:13:00.052+01:00

It's very common in the world of Java that you are developing an application using some libraries or open source frameworks or some 3rd party jar and while executing your application/program all of a sudden you get NoClassDefFoundError. This is because the runtime dependency of one of the jars you are using is not in the classpath of your application. Next step you do is start wondering where you will find that particular class. Probably you already know which jar to include and probably you have no clue where to look for that particular class. In the later case (almost all the time with me :-)) you end up hitting the icon of your favorite browser and start googling. In 99% cases you will get the name of the jar where to look for the class but in some cases, where the missing class is a part of some jar which is very much specific to some internal application or 3rd party library, google will be of little help.

I've been in such a situation many times and there are instances where I have invested (read wasted) significant amount of time just clicking through directories, opening jar files and finding the class am looking for. One fine day I realized that I should have something which does this for me automatically and easily. Result of this thought was a small utility - JAR Reader - which is having a small cute swing UI where I can

Specify which class to find
In which directory to find
Whether to look for the exact match or just part of. This is useful when I'm just looking for a package name or when I just know the name of the class and not sure of package name.

Download JAR Reader

This small handy utility has save a lot of time of mine. It searches for class in all the available jars under one directory and all the sub-directories within the search root directory. I'm sure there are many out there who would be facing same problem and for all those poor souls, I've made this download link available to download the utility. Download it and use it however and when ever you want to use it.

To start the JAR reader utility, just open the command prompt, change the directory where you have copied the jarreader.jar file and execute following command...

java -jar jarreader.jar

with above you should be seeing as swing window as the snapshot above...

@Also See: Online site which can tell you the name of the jar file for a particular class. A nice bookmark to have with you all the time. http://www.jarfinder.com/

Cheers !!!
- Jay

Problem with OJDBC 14, PreparedStatement & Statement batch execution

2009-10-04T16:51:00.015+02:00

Recently I was working on an application where we are using iBatis data persistence framework with Oracle 10g and Java 1.4.2. This application basically plays with considerably huge amount of data which it reads from some file over the queues. Just few days back we received a huge file with around 66976 records. Application reads/parses all records, does some logic processing and then inserts all records in database as a JDBC batch through iBatis. To our surprise it was just inserting 1446 records and returned silently. It was behaving like everything went well and all records have been inserted.

After debugging a lot (for 2 days) at 2 in the morning we came to know that application's business logic is just working fine and is not gobbling up records due to some bug. Its the iBatis call to executeBatch was creating problem. All of a sudden one of the team members - Chaitanya(@chaituvj) - realized that missing number of records are 65536 (66976 - 1446 ) i.e 2^16. He had haunch that somehow iBatis is not able to handle a batch with more than 65536 records. Upon doing a small PoC this turned out to be true. In PoC we also wrote a small pure JDBC code where we created a batch of java.sql.Statement with 66000 records and tried to insert it. It just went fine. Further debugging iBatis code we came to know that iBatis internally calls java.sql.PreparedStatement.executeBatch() and till that point everything is fine. No records are lost. We almost came to a conclusion that iBatis is the culprit. To enforce our finding, we extended our PoC to have batch execution with MySQL to insert more than 65536 records with iBatis and pure JDBC calls. It went fine. No issues. So once again our understanding was reinforced by this PoC result - iBatis is the culprint.

After one more day of brain storming we noticed that in our pure JDBC code we had a batch of java.sql.Statement and iBatis is using java.sql.PreparedStatement. Since PreparedStatement extends Statement, there was no reason to doubt there might be any problem with PreparedStatement's executeBatch() method implementation. But still there was no satisfactory resolution to this strange behavior, we thought to give it a try and changed our pure JDBC PoC to use java.sql.PreparedStatement instead of java.sql.Statement and execute the program to insert 66000 records. To our surprise, it inserted only 464 records i.e. 66000 - 65536. So now we found our real culprit - Oracle implementation of java.sql.PreparedStatement with ojdbc14.jar

This is a real strange implementation of PreparedStatement's from Oracle's ojdbc driver. If Oracle guys are reading this hope to hear some nice explanation for such implementation. ;-)

[Update: 3rd Dec 09]: Posted the same issue on OTN forum and found that its a bug with oracle driver version 10.1.x. Check this link for more details.

Cheers !!!
- Jay

Exporting eclipse plugins locally from one instance to another

2009-08-13T07:14:00.022+02:00

As an ardent fan of Eclipse IDE I always try to use the latest available release to see what all new cookies are there for me.

With this the only painful thing I've faced till now is every time I've to install/download all of my plugins like Maven, SpringIDE, Subversive etc. Many provides local archive installation and many needs to be updated every time using Eclipse. This is very time consuming and also some times not possible in cases where I do not have internet connectivity for what so ever reason.

Recently I downloaded Galileo and again faced the same problem as I described above and this time I discovered that Eclipse has an inbuilt feature to import plugins from any of the existing eclipse install to new install without need to go to update site or download it from the internet. Here I'll display how to do this...

Download latest Eclipse from download page and start the IDE.
Setup a local repository to point to your old-eclipse-install's plugin download directory. Eclipse stores all downloaded plugins to <eclipse-install-directory>/p2 directory. Below are the steps to setup new local repository with Galileo
1. Select Install New Software... from Help menu
2. Click Add... button to add new repository. See below snapshot...
3. Click on Local... button and select the p2 directory of your existing eclipse install. For e.g. if your existing eclipse is installed at C:/eclipse then browse for folder C:/eclipse/p2
4. Provide appropriate name of your choice for this new repository and click OK.
5. You should see the list of plugins available in your existing eclipse install. In case you do not see it then uncheck Group items by category checkbox. See below snapshot...
Once your local repository is set, select all the plugins you want in your new install and click Next and then Finish.
This will calculate the dependencies and then install the selected plugins from your existing/old install.
Done.

Cheers !!!
- Jay

Static method access using OGNL & Struts 2.1.x

2009-07-02T15:00:00.005+02:00

We know that if we want to access any static method using OGNL in Struts 2 we use notation @classname@methodname( args ) and it works well. With Struts 2.1.x, by default this has been turned off due to some security issues. You can still access static methods but, you need to tell struts explicitly to allow access to static methods. To do this in you struts.properties put following line…

struts.ognl.allowStaticMethodAccess=true

For more details on this change check Jira WW-2160

I wasted an hour on figuring out why my static method is not called and found this fix with Struts 2.1.x. If you guys are working on 2.1.x then keep this in mind in order to stay away from frustration ;-).

Cheers !!!
- Jay

P.S.: Actually I should have read Struts 2.0.x to Struts 2.1.x migration guide before starting working on Struts 2.1.x

ATM Security Myth ------- Busted

2009-06-24T07:58:00.020+02:00

Many of us have received a mail saying what to do when a thief/mugger traps you at a ATM. Below is one such mail I received couple of days back...

WHEN A THIEF FORCES YOU TO TAKE MONEY FROM THE ATM, DO NOT ARGUE OR RESIST, YOU MIGHT NOT KNOW WHAT HE OR SHE MIGHT DO TO YOU. WHAT YOU SHOULD DO IS TO PUNCH YOUR PIN IN THE REVERSE, I..E IF YOUR PIN IS 1254, YOU PUNCH 4521.

THE MOMENT YOU PUNCH IN THE REVERSE, THE MONEY WILL COME OUT BUT WILL BE STUCK INTO THE MACHINE HALF WAY OUT AND IT WILL ALERT THE POLICE WITHOUT THE NOTICE OF THE THIEF. EVERY ATM HAS IT; IT IS SPECIALLY MADE TO SIGNIFY DANGER AND HELP. NOT EVERYONE IS AWARE OF THIS.

FORWARD THIS TO ALL YOUR LOVED ONES, FRIENDS AND THOSE YOU CARE

I instantly forwarded this mail to HDFC and ICICI banks and below is the response which I received from the HDFC bank...

Dear Mr. Khimani,

Thank you for writing to us.

This is with reference to your e-mail dated June 16, 2009 regarding Security issues in ATM.

Currently, this facility is not available with HDFC Bank. We request you not to try using your ATM PIN in wrong way. It might block your ATM transaction.

And below is the response from ICICI bank...

Dear Customer,

Thank you for writing to us and bringing the e-mail below to our notice.

We affirm that keying a PIN in reverse order into our ATMs will not yield the result described in the e-mail that you have quoted.

It will yield an “Error” message on the ATM screen.

Thank you for the opportunity of issuing this clarification.

Looking forward to more opportunities to be of service to you.

Sincerely,
Rose
Customer Service Manager

Such facility is not only NOT available with an Indian bank, but no bank ATMs in world provide such reverse pin feature. Myth busters Barbara and David Mikkelson of snopes.com has also turned such information (read rumor/spam) as false. For more details check this article on snopes. http://www.snopes.com/business/bank/pinalert.asp

So friends, please DO NOT try to display any such smartness until you are sure whether the ATM you are using is providing such facility. Instead try to be as cautious as possible while entering any ATM and try to avoid any ATM which is in very remote location and you do not see anybody around.

Cheers !!!
- Jay

Firefox 3.5 & Google Notebook extension

2009-06-11T08:14:00.009+02:00

Being an ardent Firefox fan, I could not wait to download Firefox 3.5 beta and see how fast it is and what all new features are added. After installing it over 3.0.10 and restarted in anticipation to see something new, something cool, surprisingly I saw a message that my favorite google notebook extension is no more compatible with Firefox 3.5. With utter despair, all thoughts of exploring new Firefox vanished and I started googling on how can I get my notebook back. Didn't had to wait long. Found this quite useful blog which made me happy again with no complaints for Firefox. ;-)

http://googlesystem.blogspot.com/2009/06/google-notebooks-extension-url.html

Follow the instruction on above and you can get have your google notebook extension back.

Addition on July 03, 2009 18:41:29
If you are too lazy to do all the above then Steve is your savior. Visit http://www.notebookaddon.tk/ and you can install modified google notebook extension without all the jugglery. Thanks Steve.

Cheers !!!
- Jay

P.S.: Above blog also introduced me to a good free archiving utility 7-zip. No annoying winzip evaluation dialogue or winrar's purchase license dialogue. ;-)