Schedule Job automatically

[Vishvas]

We have more than 30 schedules in our project. In case of any downtime we haev to reschedule all the 30 schedules manually. This is time consuming and error prone. So I was thinking of some solution where we can have the schedules in a excel and a job will read the excel and change the schedules accordingly.

I know that Datastage uses UNIX scheduler for scheduling. crontab & at will give you the info about all the schedules. But if you change the crontab / at schedule at the backend it doesn't affect the schedule in the DataStage Director.

Is there anyway to build such a job for rescheduling.

Arun

[ray.wurlod]

No.

But, if you capture the queued commands for cron and/or at while things are good, you can resubmit directly onto the queues.

[Vishvas]

The only issue is that it won't be reflected in the director. The director will show a different schedule and the job will run at different schedule.

[chulett]

That's not true... all the Director does is read the crontab entries for the logged in user and show any of the "DataStage ones" that it finds. Any proper change on the underside should show just fine, AFAIK.

[kwwilliams]

The only issue is that it won't be reflected in the director. The director will show a different schedule and the job will run at different schedule.

Most people wouldn't use Director as the scheduling tool. Director through cron is a time based tool that cannot take into account the dependencies usually needed to build out a warehouse or any other system. This would cause you to build out super-sequence jobs, essentially putting every job and every dependency within a sequence. This would ensure that no job starts before its upstream dependency is complete, but is very rigid in structure. The rigidity will cause problems in the future if you need to rework your dependent objects based upon new requirements or peaks in your hardware utilization.

If you insist on timebased scheduling, I would ask if you have a scheduling tool in-house such as CTRL-M or Active Batch. These tools have the functionality built into them that you desire. IBM is not investing in the scheduler, so I wouldn't expect to see any improvement in the functionality in the future.

Just curious if you have thought about a pure dependency based scheduling model, without the time constraints of a scheduling tool. You would essentially track what is needed for a job to start, and when all of the upstreams dependencies kick off the next job.

[chulett]

Well... that's a whole 'nuther discussion and really outside of the scope of what was being asked here. All pretty much true, however.

[kwwilliams]

Well... that's a whole 'nuther discussion and really outside of the scope of what was being asked here. All pretty much true, however.

True. I could have answered the question without opening another discussion. However, when people are trying to make a square peg fit in a round hole I want them to know it.

[chulett]

... and one man's round hole is another man's square hole. Or something like that. LOL

[Vishvas]

No, Datastage Director doesn't reflect the changes to the timings made at the backend. It takes the details from RT_CONFIGnnn file for displaying. But if you delete a schedule from backend, that will be reflected in the Director.

[chulett]

That's... interesting. Too bad I don't have any way to play around with or test anything like that.