Spike: Offer simple ORM interface for scheduling a job

[lampwins]

Environment

• Python version: 3.8.11
• Nautobot version: 1.2.4

Proposed Functionality

As P.D. the plugin developer, I want a simple interface to schedule a job execution via the ORM so that I can easily perform this action within the business logic of my plugin.

Use Case

Today it takes ~30 lines of code to create a ScheduledJob instance for a nautobot job. Example:

nautobot/nautobot/extras/views.py

Lines 820 to 857 in 0e491e1

	if job_class.approval_required or schedule_type in JobExecutionType.SCHEDULE_CHOICES:
	if schedule_type in JobExecutionType.SCHEDULE_CHOICES:
	# Schedule the job instead of running it now
	schedule_name = schedule_form.cleaned_data["_schedule_name"]
	schedule_datetime = schedule_form.cleaned_data["_schedule_start_time"]
	else:
	# The job must be approved.
	# If the schedule_type is immediate, we still create the task, but mark it for approval
	# as a once in the future task with the due date set to the current time. This means
	# when approval is granted, the task is immediately due for execution.
	schedule_type = JobExecutionType.TYPE_FUTURE
	schedule_datetime = datetime.now()
	schedule_name = f"{job.name} - {schedule_datetime}"
	job_kwargs = {
	"data": job_class.serialize_data(job_form.cleaned_data),
	"request": copy_safe_request(request),
	"user": request.user.pk,
	"commit": commit,
	"name": job.class_path,
	}
	scheduled_job = ScheduledJob(
	name=schedule_name,
	task="nautobot.extras.jobs.scheduled_job_handler",
	job_class=job.class_path,
	start_time=schedule_datetime,
	description=f"Nautobot job {schedule_name} scheduled by {request.user} on {schedule_datetime}",
	kwargs=job_kwargs,
	interval=schedule_type,
	one_off=schedule_type == JobExecutionType.TYPE_FUTURE,
	user=request.user,
	approval_required=job_class.approval_required,
	)
	scheduled_job.kwargs["scheduled_job_pk"] = scheduled_job.pk
	scheduled_job.save()

In the spirit of JobResult.enqueue_job(), it should not be this hard to schedule a job for execution. We can likely consolidate the JobResult.enqueue_job() method into this too.

Should have this interface live on Job (ex: Job.schedule_job()) with the hope of collapsing JobResult.enqueue_job() onto Job somewhere in the future.

Database Changes

None

External Dependencies

None

TODO:

This should cover interfaces for both scheduled jobs and immediate execution.

[tbotnz]

@lampwins/whoever resolves this issue, would you also be able to consider a pattern where JobResult.enqueue_job() is triggered by django signals.
The current arg (user) on JobResult.enqueue_job() makes this pattern (signals triggering jobs) problematic to implement

[lampwins]

Converting this to a spike with the goal of defining what the interface for a combined enqueue/scheduling method would look like.

[gsnider2195]

@lampwins/whoever resolves this issue, would you also be able to consider a pattern where JobResult.enqueue_job() is triggered by django signals. The current arg (user) on JobResult.enqueue_job() makes this pattern (signals triggering jobs) problematic to implement

That should be covered by the work for #1631

[bryanculver]

Design complete. See #2474. To be implemented via #2535