Our client needs an restarting/rerunning mechanism in datastage. The story behind this is.. they have a number of jobs in a datastage project which runs daily. This is in version 7.5.2. The problem is that at times some jobs finishes successfully but does not flag the sequencer which invokes it and hence the sequencer sits there waiting for the child job to finish and holds up the downstream processes.Eventually, the production support guys are alarmed and they have to just rerun the sequencer and it runs fine.Another project which had the same problem earlier seems to be working now when it is a upgraded to datastage version 7.5.3.
Now, the client needs a mechanism which will work irrespective of the version of Datastage. They want an restarting mechanism which can rerun the seqencer without raising an alarm (no manual intervention at all) . What they are looking for is:
1. Restart/rerun mechanism within datastage which will just rerun the sequencer in such scenarios.
2. If the job fails for some other genuine reason then it should not be restarted/rerun but should raise the appropriate alarms.
3. The mechanism should be intelligent to make only 2-3 attempts to rerun it and should not keep doing it for ever.
4. The source for these jobs are files which are picked up from one folder in unix and after job run they are archived to an archive folder.So the Datastage solution which we intend to implement should have a logic to copy the files back from the acrhive to the actual folder before rerun.
I am not quite sure whether all these can be implemented just through datastage alone.
Please provide your valuable suggestions.Note: I have gone through other posts asking for similar solutions. But I guess all the solutions had some manual work involved, but in my case the solution I am suppposed to design
should be such that there is ZERO manual work.