PDG/Deadline -> Fail Task = Endless Loop

   498   0   1
User Avatar
Member
10 posts
Joined: June 2022
Online
Hello,

The short of it -
How do I override a Deadline Job's Properties for failure detection from the Deadline Scheduler?
Deadline Failure Detection Docs [docs.thinkboxsoftware.com]


Long of it -
I'm submitting PDG Python Script Work Items to Deadline via the DeadlineScheduler. Each Work Item gets it's own job, containing a single task within (not using "Submit As Job" option, just cooking out-of-process tasks via DeadlineScheduler)

There is a condition in the Python Script that fails the task -
work_item.addError('Fail Note', fail_task = True)

After the condition is hit, the error is thrown, the task fails, HOWEVER... the same task re-queues itself in Deadline and does not fail the job. The work_item.addError condition is met again, task fails, task requeues, endless loop.

How do I get the deadline JOB to fail if the addError condition is met?

Looking at the Manual Job Submission [docs.thinkboxsoftware.com] options available, I see a few for "Failure Detection" -


I've attempted settings these job parameters in multiple places around the Deadline Scheduler and they do not pickup in Deadline -



So how do I go about failing the Job when the single job's task fails via work_item.addError(fail_task=True) and not have just an endless loop of the same failed task requeuing itself?

Thank you for any help / input.
Edited by krause_trk - Sept. 24, 2024 17:06:45

Attachments:
deadline.jpg (93.0 KB)
failure_detection.jpg (178.3 KB)
job_parms.jpg (113.7 KB)
job_spec2.jpg (98.8 KB)

  • Quick Links