Restartability while running multiple Datastage jobs through

Mayur Dongaonkar · Post by **Mayur Dongaonkar** » Wed Nov 28, 2007 5:03 am

My requirement is to call three datastage jobs through a unix shell script. Shell script will call one job after another.
But if any job aborts, then during re-run shell script should start running the jobs from point of failure.

Example:
Shell script is calling Datastae jobs

A -> B -> C

If A completed successfully and B got aborted then during re-run shell script should run only jobs B and C.
We really do not want to create any temporary file on Unix to store status of jobs.

Could you please help me.

Maveric · Post by **Maveric** » Wed Nov 28, 2007 5:21 am

You can do this using a job sequence. Any specific reason to use shell script?

Mayur Dongaonkar · Post by **Mayur Dongaonkar** » Wed Nov 28, 2007 6:01 am

Our scheduler will call only unix scripts.

Maveric · Post by **Maveric** » Wed Nov 28, 2007 6:10 am

Build a sequence which takes care of restartability and trigger it using a unix script.

Mayur Dongaonkar · Post by **Mayur Dongaonkar** » Wed Nov 28, 2007 6:40 am

Sequencer will take care for restartablity. I agree with it. But we really do not want to build additional sequencer for this requirement. We want to keep it between single Unix shell script and multiple Datastage jobs in it.

gateleys · Post by **gateleys** » Wed Nov 28, 2007 7:59 am

Mayur Dongaonkar wrote:We want to keep it between single Unix shell script and multiple Datastage jobs in it.

Why not keep it between a single Master Sequence and multiple jobs in it? The only thing that you will do from UNIX is to call dsjob to execute the master sequence. This way you can use your enterprise scheduler, and at the same time, have a single entity (Master Sequence) to monitor and manage.

Instead of spending time on changing your UNIX script, why not allocate that time to building a restartable sequence? Its more maintainable, and I assure you it won't take long.

But then, that's how I see it, others may wish to differ.

chulett · Post by **chulett** » Wed Nov 28, 2007 8:13 am

Agreed. It makes zero sense to try and recreate all that restart logic in a shell script when all you need to do is create a Sequence job that will handle all that automatically and then run the Sequence job via the script.

ray.wurlod · Post by **ray.wurlod** » Wed Nov 28, 2007 2:51 pm

Mayur Dongaonkar wrote:We want to keep it between single Unix shell script and multiple Datastage jobs in it.

Then you're on your own. DataStage provides a tool: you choose not to use it.