Sequence - Best Practice

mchaves · Post by **mchaves** » Sat Jan 14, 2006 6:23 pm

Hi there,

Another question to the online gurus.

I would like know the best practices when setting up a sequence of jobs. I have questions (not to say frustrations) about how to build a robust sequence. Those are the main questions:

1- How to setup a sequence that can restart at the point of failure after having the cause of failure resolved?

2-Should I use the gui to build a robust sequence or a control job?

3-How to set priorites to the jobs which allow the sequence to decide whether to go on or to abort according to importance of the job.

Please, add any information about robust sequences I forgot to ask which will enrich this topic.

Cheers,
MC

vmcburney · Post by **vmcburney** » Sat Jan 14, 2006 6:32 pm

If you are using the latest version (7.5.1) you can use the automated checkpoint restart built into sequence jobs. Go to job properties and look at the check boxes on the first properties screen. When a job aborts you will see it has the status of aborted/restartable, this means you can restart it and it will start from the job or jobs that failed.

Have a look at the FAQ on project specific environment variables for a method of setting variables in Administrator and using them in a sequence job.

For slowly changing variables such as file directories and database connection parameters you can set them through Administrator and update them when required. For dynamic variables such as processing dates and surrogate key starting values you can set them via the Sequence job Variables stage which gives you easy access to BASIC routines and external scripts for retrieving values. These values can then be passed to any subsequenct stages as normal parameters.

kumar_s · Post by **kumar_s** » Sat Jan 14, 2006 9:06 pm

Hi,

Once i faced an issue with the restart capability in sequence.
For the jobs we have an option to reset if required and restard. So that even if the job is abored, during the rerun it can be restared.
But the same way subsequence cannot be restarted.
Since we dont have that option for subsequence, by any chance if subsequence got failed, and if happend to reset the main sequence and restart from the first, we cannot expect the the subsequence to start from the first, rather it will just restart from where its previous checkpoint is marked.

Should make sure the subsequence also made to reset.

-Kumar

vmcburney · Post by **vmcburney** » Mon Jan 16, 2006 12:49 am

There is certainly a support training requirement, if a sequence job is at the status aborted/restartable then simply starting it will run from the failed checkpoint while resetting it will clear checkpoints. Support staff need to know when to run and when to reset. Simply starting a top level sequence without knowing what the statuses of the next level of sequence jobs are can be dangerous.