Wierd issue with DataStage job-Job hangs for 2nd run onwards

[hcdiss]

I have a Datastage job which reads just 500 records pivots each record into 100 records and later joins those 50000 records with other tables to make final count 300,000.
The wierd problem is when I compile the job and run it, it takes 3 minutes to complete. If I rerun the same job it hangs forever. I have to abort it after few hours. If I compile the job again, it runs within 3 minuts.
Any one has any idea about this issue? Does anyone know why this could be happening?

[bobyon]

Check with a DBA and see if there are deadlocks occurring in the table. Maybe something is not ending properly and leaving lock(s) in place.

[hcdiss]

The job just reads from a table and writes to a dataset. I'm not writing data to output tables.

[SURA]

1) Which version of Datastage are you using?

2) Did you applied the latest fix for MKS Tool kit?

3) When you run the job ask your sys admin / if you can see the event log and find is there is any Nu Tracker errors are happening while running the job.


4) When you run this job, this is the only job running / any other jobs running?

5) Ensure that you handled the data type conversion (If any) properly! In Windows if the Developer fails to handle such situation & If you don't have the latest fix for MKS Tool kit, then MKS won't able to handle it and end up with NuTracker errors.

Some of the NuTracker error will keep on continue until the server is restarted. This will affect the total DS performance too!

6) Add environment variables like score dump, there is an another variable (Sorry i couldn't remember ) to capture the job start-up details and other stuff!

So lot of ways you can find where the issues could be.

Start with one after other and narrow down the issue one by one.