Manually triggering a job - trigger dependent jobs also?

[Kasper Sørensen]

Hi all,

I'm currently looking at the alpha version of DataCleaner 3.0 and wondering about the "Trigger now" button available in the scheduling panel.

This button triggers the manual execution of a single job.

My question is: What if there are other jobs that are depending on this job (aka. cascading execution)? Either 1) the dependent jobs will also be executed, or 2) the "Trigger now" button will only trigger the single job.

Going for the first option would mean that "Trigger now" could potentially trigger a lot of jobs, if there is a large dependency graph.

Going for the latter option would mean that cascaded execution could only be triggered by some other means (ie. an originating periodic trigger).

What would be best? Looking at/comparing with eg. typical Continuous Integration systems like Jenkins, I think maybe the first option (trigger also dependent jobs) is best?

[Pulin Pathneja]

If we don't trigger other dependent jobs, won't we have corrupted or obsolete data?

[Kasper Sørensen]

That is a possibility I guess, in a kind of ETL-like scenario. I was just wondering if some users might see it the other way around, thinking "if I manually trigger something, then the application should not (automatically) trigger anything else".

But maybe that's just speculating too much. Having the cascade effect seems a bit more consistent I guess.

Other perspectives?

2012/8/21 Pulin Pathneja <pulin.p...@gmail.com>

--
You received this message because you are subscribed to the Google Groups "DataCleaner-dev" group.

To view this discussion on the web visit https://groups.google.com/d/msg/datacleaner-dev/-/KUZ-WB7yGmMJ.

To post to this group, send email to datacle...@googlegroups.com.
To unsubscribe from this group, send email to datacleaner-d...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/datacleaner-dev?hl=en.

[swapnil dubey]

I think we can follow this strategy:-

When we manually trigger a job, the other jobs dependent on this do not run immediately but at a fixed time, like once in a day or hour....depending upon the nature and frequency of updates required.These time should be configurable.

[Nayan]

What i feel as a user perspectives is that it would be great if we ask user itself for triggering option. 

Something like following scenario 

when user manually trigger a job then all the dependent jobs can be populated in form of checkboxes and from there user can select which dependent job to run along with main job.

[Kasper Sørensen]

I agree this would be the optimal solution. But it might also make things more difficult and since I think we need to focus on getting things ready for a release, I propose that we start with just a basic feature. I think the concensus seemed to be that cascading jobs should be fired by default, so I've created a trac ticket for that:  http://eobjects.org/trac/ticket/924 

2012/8/21 Nayan <nayan...@gmail.com>

To view this discussion on the web visit https://groups.google.com/d/msg/datacleaner-dev/-/W6aD_LfeVWsJ.