How to drink a beer and sleep well while your Apex Jobs are running?

Catching Job Errors and Re-firing them using Apex Scheduler

Pain, pain, pain, oh sometimes it pains in the ass when no matter what you do, no matter how much you brain it out and architect the delicate pillars to build a huge Apex batchable job that processes tons and tons of records, there are many times your batch job might fail. And that failure could throw you out of your job if your batch job fails to execute an ultra-important business calculation that the business folks were relying on. The damn bloody error could be because of a CPU timeout, a concurrency lock error, or a Salesforce Internal Server Error due to a SOQL query that is taking more time to execute than allowed by the god damn governor limits. And you cant be bloody monitoring the job waking up in the middle of the night, to re-fire it manually when you realize it didn't fire the automated way. Sometimes it might even happen that the error handler in your code's Exception block might not fire because of an Internal Server Error. And you were expecting like an idiot an email to your inbox that would tell you what's the status of the job - Completed or Aborted? 

So what can you do to handle such crazy scenarios?

Well the following approach is a bit of a work-around, until you manage to build a Cadillac out of your code, simply check for errors and re-fire your batch job if it didn't run successfully in the last fire.

Now here comes to rescue the super duper AsyncApexJob sObject. Simply query it, ask it if your darling job fired. To start with, I must say I love this object for many reasons that I can keep on going on.

Enough of shit stories, here is the approach:
* Query the AsyncApexJob object to find out the status of the last fired instance of your batch class
* If the job didn't run happily, just refire your job.

Now say the name of your batchable job is MyDearBatchable. And you or some external source like an ETL fired MyDearBatchable to run. But then due to some stupid bloody reason, MyDearBatchable didn't run successfully.

So what you do is create a Schedulable job like the one below:

global class MyDearSchedulable implements Schedulable{
    global void execute(SchedulableContext ctx) {
        // Fetch status of the MyDearBatchable job that ran last
        List<AsyncApexJob> listStatusOfBatch = [SELECT Id, MethodName, JobItemsProcessed, ApexClassId, ApexClass.Name, CompletedDate, NumberOfErrors, Status, ExtendedStatus, TotalJobItems FROM AsyncApexJob where ApexClass.Name='MyDearBatchable' order by CompletedDate Desc Limit 1 ];
        
       Boolean refire1 = false; // status flag for refiring IntradayEmailBatchable
        for(AsyncApexJob j1:listStatusOfBatch){
            if(j1.Status=='Aborted'){
              refire1 = true; 
            }
         }
        
        // Refire MyDearBatchable 
        if((refire1==true){
            Database.executeBatch(new MyDearBatchable(null), 200);
        }
    }

}

Now you can schedule this class to run by navigating to Setup -> Apex Classes and then click on "Schedule Apex Jobs", or alternatively, schedule it to run using the Developer Console using the following command:

MyDearSchedulable m = new MyDearSchedulable(); 

string sch = '0 20 10,14,16 * 1,2,3,4,5,6,7,8,9,10,11,12 ? 2015'; // You can set your own CronExpression here to set the time you want it to fire in Queue 

system.schedule('MyDearSchedulable', sch, m);

And thats it. You can sleep happily ever after. You just need to know & guess when to schedule this job.

Have a beer & sleep well.

Cheers!
Vishal Khanna