Zip file download problem with IE

There is a zip file download button in our application for sometime. Recently to make the browsers identify the file, we set the content type as "application/zip". Which is correct according to the standard[1]. But we found that the file downloaded with IE seem to be corrupted. While trying to investigate the issue, we were absolutely clueless on how to go about this problem. FF and some download managers with IE also seem to be downloading the file properly! After quite some frustrating hours of debugging we replaced the content type with generic: "application/octet-stream". And IE seem to understand this finally.

1. http://www.iana.org/assignments/media-types/application/zip

Configuring LAN with BSNL Datone in Linux machine

This is how I configured my home LAN with one desktop and one laptop. Before listing the exact steps let me brief the background. The desktop runs on CentOS and acts as the gateway for the LAN machines. Main difficulty was implementing the NAT so that I could access Internet from my Laptop. Looking at the tldp documentation, I thought it would be fairly easy to implement ip masquerading using iptables. Actually it was but only after I found the correct fix when Internet is provided through PPPoE protocol. The point to remember with PPPoE is the outgoing port in these situation will be ppp* not eth*.

     +-------------+               +-------------+               +-------------+
 ----|             |               |             |  10.0.0.*     |             |
 BSNL|  Modem      |  192.168.1.*  |  Gateway    |---------------| Internal M/C|
Cable|             |---------------|  (CentOS)   |eth1           |   (Ubuntu)  |
     |             |    (ppp)  eth0|             |               |             |
     +-------------+               +-------------+               +-------------+

Step 1:

[root ]# echo 1 >  /proc/sys/net/ipv4/ip_forward

Step 2:

[root ]# echo "net.ipv4.ip_forward = 1" >> /etc/sysctl.conf

Step 3:

[root ]# echo 1 >  /proc/sys/net/ipv4/ip_dynaddr

Step 4:

[root ]# /sbin/iptables -A FORWARD -i eth1 -o ppp0 -m state --state RELATED,ESTABLISHED -j ACCEPT

Step 5:

[root ]# /sbin/iptables -A FORWARD -i ppp0 -o eth1 -j ACCEPT

Step 6:

[root ]# /sbin/iptables -t nat -A POSTROUTING -o eth1 -j MASQUERADE

Step 7:

[root ]# service iptables save

Step 8:

[root ]# service iptables restart

Step 9:
If you have some firewalls (like Redhat Firewall) enabled in your system, you must make the LAN interface trusted in the gateway machine.

And there you go. Step 4-9 can also be done using UI: /usr/bin/system-config-securitylevel-tui like this:



P.S.: Chennai LUG helped me for this.

One CVS trick with eclipse

Recently I had a need to commit some local uncommitted change of a branch into another branch. I was little skeptical that I might loose all the local changes and redo them, manually, again in the other branch. Some soul searching in eclipse help document, revealed that I can do such things using "Team -> Switch to another branch or version". It uses sticky tags to move the uncommitted changes into the new version/branch instead of replacing. Once you have "switched" to this new branch, you can commit them gladly. And, it works! But there is one problem. Suppose there is some conflicting file, i.e., you have modified a file "A" locally and someone else also has made some conflicting changes in "A" in this other branch. In this particular situation, eclipse/cvs screws your code gracefully. i.e., without any warning/error. It looses all local changes you made! So my personal opinion is before "switching" to another branch, do a "compare with" that branch. Find out if there is any conflict. In case you find any conflict, back them up somewhere. After "switching" to the other branch manually check which state these files are. You might need them to fix manually.

Quick HOWTO : Quartz with JBoss

While searching for some job scheduler tool, I came across this Open Source tool, called, Quartz. Most importantly my requirement was to be able to use it from inside an application server like JBoss. After reading some tutorials, I gave it a go. Let me just brief my experience here.

Quick Intro
Fundamental entities with Quartz or with any other job schedulers are tasks and triggers invoking them. Quartz has got two kinds of triggers: SimpleTriggers and CronTriggers which are very much similar to Unix Cron triggers. There are two ways one can store the jobs in Quartz. RAMJobStore is the most simple one but the tasks stored here do not get persisted, i.e., jobs get forgotten once the server restarts. Other type of job stores are JDBCJobStore. As the name suggests with these job stores, jobs are stored in relational databases. Quartz supports most of the standard databases like MySQL, Oracle and etc. There are two kinds of JDBCJobStores available: JobStoreTX and JobStoreCMT. JobStoreTX is supposed to be used stand alone whereas JobStoreCMT is meant to be used from an application server. As my requirement was to use it with JBoss I used only JobStoreCMT. And, used it with MySQL as the back end database.

Quartz as a JBoss service
JBoss 4.2.2.GA already has got quartz bundled with it. But actually you can not do much with this as it supports only RAMJobStore. So to use it effectively we need to do some more. We have to copy quartz-1.6.2.jar and quartz-jboss-1.6.2.jar in the JBOSS_HOME/server/server_mode/lib directory. Then, we need to create a xml file to configure Quartz as a service in JBOSS_HOME/server/server_mode/deploy directory.

Create the data base
We need to create the database where all job and trigger information will get stored eventually. The sql scripts are bundled with Quartz download under docs/dbTables directory.

quartz-service.xml
As already mentioned we need to create the quartz-service file . Let me just enclose a sample file.

<server>

<mbean code="org.quartz.ee.jmx.jboss.QuartzService" name="user:service=QuartzService,name=QuartzService">
<!--       JNDI name for locating Scheduler, "Quartz" is default.     -->

<attribute name="JndiName">Quartz</attribute>
<attribute name="Properties">
# Default Properties file for use by StdSchedulerFactory
# to create a Quartz Scheduler Instance, if a different
# properties file is not explicitly specified.
#

# org.quartz.scheduler.classLoadHelper.class =

org.quartz.scheduler.instanceName = DefaultQuartzScheduler
org.quartz.scheduler.rmi.export = false
org.quartz.scheduler.rmi.proxy = false
org.quartz.scheduler.xaTransacted = false

org.quartz.threadPool.class = org.quartz.simpl.SimpleThreadPool
org.quartz.threadPool.threadCount = 5
org.quartz.threadPool.threadPriority = 4

org.quartz.jobStore.class = org.quartz.impl.jdbcjobstore.JobStoreCMT
org.quartz.jobStore.driverDelegateClass =  org.quartz.impl.jdbcjobstore.StdJDBCDelegate
org.quartz.jobStore.dataSource = QUARTZ
org.quartz.jobStore.nonManagedTXDataSource = QUARTZ_NO_TX
org.quartz.jobStore.tablePrefix = QRTZ_
org.quartz.dataSource.QUARTZ.jndiURL = java:QuartzDS
org.quartz.dataSource.QUARTZ_NO_TX.driver = com.mysql.jdbc.Driver
org.quartz.dataSource.QUARTZ_NO_TX.URL = jdbc:mysql://localhost:3306/Quartz
org.quartz.dataSource.QUARTZ_NO_TX.user = root
#org.quartz.jobStore.maxMisfiresToHandleAtATime=0

</attribute>

</mbean>
</server>

The point to note here is that we'll need two datasource elements for this configuration. One is standard datasource managed by JBoss container (see the line org.quartz.jobStore.dataSource = QUARTZ) another one is not managed by the container, on which quiatz can call commit/rollback by itself. We have to confugure the container managed datasource as a standard jboss *-ds file, whereas we configure the non_managed_datasource inside this quartz-service.xml itself. Here one gotcha is for the container_managed datasource, we'll need to use XA datasource (I am not very sure why regular local datasource does not work. Maybe reason being Quartz works in a clustered environment, just a guess though). Let me just attach a sample mysqlquartz-ds file here:

<?xml version="1.0" encoding="UTF-8"?>
<datasources>
<xa-datasource>
   <jndi-name>QuartzDS</jndi-name>

   <xa-datasource-class>com.mysql.jdbc.jdbc2.optional.MysqlXADataSource</xa-datasource-class>
   <xa-datasource-property name="URL">jdbc:mysql://localhost:3306/Quartz</xa-datasource-property>
   <user-name>root</user-name>
   <password></password>
   <transaction-isolation>TRANSACTION_READ_COMMITTED</transaction-isolation>
   <max-pool-size>5</max-pool-size>
   <min-pool-size>0</min-pool-size>

   <blocking-timeout-millis>2000</blocking-timeout-millis>
   <idle-timeout-minutes>2</idle-timeout-minutes>
   <track-connection-by-tx>true</track-connection-by-tx>
   <no-tx-separate-pools>false</no-tx-separate-pools>


   <exception-sorter-class-name>org.jboss.resource.adapter.jdbc.vendor.MySQLExceptionSorter</exception-sorter-class-name>
<!-- <valid-connection-checker-class-name>org.jboss.resource.adapter.jdbc.vendor.MySQLValidConnectionChecker</valid-connection-checker-class-name> -->
   <metadata>
           <type-mapping>mySQL</type-mapping>
   </metadata>
</xa-datasource>

</datasources>

Creating a job
Once you get the configuration right, other things are pretty simple.

        InitialContext ctx = new InitialContext();
        Scheduler scheduler = (Scheduler) ctx.lookup("Quartz");
        Trigger trigger = TriggerUtils.makeDailyTrigger("myTrigger", 0, 0); //a trigger which gets fired on each midnight
        trigger.setStartTime(new Date());

        JobDetail job = new JobDetail("jobName", "jobGroup", Executor.class);

        job.getJobDataMap().put("Name", "Abdul Sahid Khan");
        job.getJobDataMap().put("Age", 125);

        scheduler.scheduleJob(job, trigger);

So you get a handle to the Scheduler object using jndi lookup. Remember we set the jndi name in quartz-service.xml file. Then we create a trigger. There is various ways to create a trigger. TriggerUtils is a utils class provided by Quartz itself which gives mane factory methods to create simple triggers. Then we need to create a job. There is no Job class (well, there is an interface by that name but let us now pretend that it does not exist), instead we will create a JobDetail object. While creating JobDetail object we need to provide an Executor class which will be used when this particular job gets fired by the Scheduler. We can store job specific information in the datamap provided by JobDetail object. These information can be used in the Executor class.

Execute the job
The job gets executed by the class which was passed while creating the JobDetail object. This Executor class needs to implement earlier_ignored Job interface. And that mandates Executor to have a method called execute().

public class Executor implements Job {
  public void execute(JobExecutionContext context) throws JobExecutionException {
      String jobName = context.getJobDetail().getName();
      String groupName = context.getJobDetail().getGroup();
      JobDataMap dataMap = context.getJobDetail().getJobDataMap();
      String name = (String) dataMap.get("Name");
      Integer age = (Integer) dataMap.get("Age");
      logger.info("Job received name: " + jobName + " , Group: " + groupName);
      logger.info("Name: " + name + " , Age: " + age);
  }
}

Delete the job
To delete a job we need to know the name and group of the job.

        InitialContext ctx = new InitialContext();
        Scheduler scheduler = (Scheduler) ctx.lookup("Quartz");
        scheduler.deleteJob("jobName", "jobGroup");

Here one must remember that you can not delete a job from inside the execute method, that will create a lock which will cause JVM to crash. So delete should be handled as different process like create.

One last thing about Jboss
Sometime, business logic needs you to have your Executor class inside the your EAR/WAR application. Now that will create a dependency problem while starting the jboss server. Each time Quartz service starts, it tries to recover if there is any misfired jobs. But if Quartz starts before EAR/WAR deploys then we have ClassNotFound problem. To solve this we can use the deploy.last hack given by JBoss. You can create deploy.last directory and put the quartz-service.xml file inside that. That will ensure Quartz service will start after all the application gets deployed.

Reference:
Quartz wiki

United we stand, do we really?

Wednesday I came back late after attending one of my friends marriage party. I switched on the TV for couple of minutes before going to bed perhaps only to discover that there had been another terrorist attack in Mumbai. Since then I feel extremely disturbed and helpless. And, being a Muslim myself, I live in this peculiar mental state of multi-directional feeling of insecurity.

Tonight I was watching this program at NDTV by Burkha Dutt "We the People". And what I saw, pretty much summed up the conflicting, ever_contrasting mood of the whole nation. In one point one of the guest, Simi Garewal, told something like: If you see the flags from all the slums surrounding the Oberoi and other multi-storied building, you'll notice these are neither Indian National Flag, nor they are any party flag like Congress's or BJPs. Perhaps she tried to mean these are flags of our *neighbouring country*. Anyhow, there was this sudden and furious response from one of the young audience: "Pakistan is not the problem, Islam is not the problem, people like Simi are the main problem..." I was absolutely stunned to see this live. I don't know if the person was a muslim himself. The significance of this show is, it shows very clearly how much united we Indians are!