Automic Workload Automation

Hi,

If you have two jobs in a workflow, JobA and JobB, with JobB being dependent on JobA. When you go into the properties for JobB, you get the option to setup the states in the dependency tab. Generally I would set it to All states must match and ANY_OK else Block. 

I have a requirement that JobB is dependent on JobA but not for JobA to finish to completion. JobA runs all day. JobB only has to wait for JobA to be in a running state, then it can start. Since JobA is still running, even if I leave the dependency state of JobA blank, JobB will never start. My guess that a previous job must finish (success or failure) before it's status is evaluated?

Is there a way around this?

Cheers

Ben

Hey Ben,

Yes you are correct, if they are in the same path in the workflow it will have to wait until the job is finished.

I would look into making it a parallel task and use the Task Properties -> Pre-conditions to 'check activities' on JobB. With that you can check to see if the job is in a running state.

Maybe something like this. You can use an alias as well so you know its specific to this workflow:

Regards,

Michael

There's definitely ways to handle this, although it might depend on what type of agent (windows vs unix) we're dealing with.

1. Update job A so that you're kicking off the process in the background.  In unix, this would mean appending a "&" to the end of the command.  It's a little trickier to do in Windows, but you can read up on the "start" command.
2. Replace job A with a task that checks to see if that process is running.  (You'd have to take care of actually kicking off job A through some other workflow.)  In unix, you could maybe use some sort of 'ps -ef | grep (job A process)'-type command to see if it's running.  Again, it'd be a bit trickier in windows, but probably possible.
3. Have a separate workflow that takes care of kicking off Job A at whatever interval is appropriate.  Right before kicking off Job A, create some temporary flag file that we can use to indicate that Job A is running (e.g., "job_a_is_running.txt").  Once Job A completes, delete that flag file.  Now, in your new workflow, replace Job A with a file event, configured for "Until first hit" and using whatever polling interval you like, that is looking for that flag file.
4. Replace job A with a job (or script) that does an activate_uc_object on job A (or a workflow containing job A).  Once this "replacement" task finishes, it means job A has been kicked off.  (You may still want to add a delay before running job B, just to make sure job A actually had sufficient time to start.)  Note that this is not necessarily the best solution, because successfully activating a workflow or job is not the same thing as confirming that it actually ran properly.

Hi BenSumner612127 

Was the information provided by Michael or Daryl helpful in answering/resolving your question?  Did you end up picking a method?

If either of the comment provided by Michael & Daryl helped in resolving your question, please use the "Mark Correct" button on their respective comment or let us know.

Thanks,
Luu