def test_skipping_all_operation_runs_sets_pipeline_run_to_finished(self):
operation_run = OperationRunFactory()
start_operation_run(operation_run)
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.CREATED
pipeline_run = operation_run.pipeline_run
assert pipeline_run.last_status == PipelineLifeCycle.CREATED
assert pipeline_run.statuses.count() == 1
# Create another operation run for this pipeline_run
operation_run2 = OperationRunFactory(pipeline_run=pipeline_run)
start_operation_run(operation_run2)
operation_run2.refresh_from_db()
# Stopping the first operation does not stop the pipeline
with patch('scheduler.tasks.jobs.jobs_stop.apply_async'
) as spawner_mock_stop:
skip_operation_run(operation_run)
assert spawner_mock_stop.call_count == 1
operation_run.refresh_from_db()
pipeline_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.SKIPPED
assert pipeline_run.last_status == PipelineLifeCycle.CREATED
assert pipeline_run.statuses.count() == 1
# Stopping the second operation stops the pipeline
with patch('scheduler.tasks.jobs.jobs_stop.apply_async'
) as spawner_mock_stop:
skip_operation_run(operation_run2)
assert spawner_mock_stop.call_count == 1
pipeline_run.refresh_from_db()
assert pipeline_run.last_status == PipelineLifeCycle.DONE
assert pipeline_run.statuses.count() == 2
def test_running_operation_run_sets_pipeline_run_to_running(self):
operation_run = OperationRunFactory()
start_operation_run(operation_run)
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.CREATED
pipeline_run = operation_run.pipeline_run
assert pipeline_run.last_status == PipelineLifeCycle.CREATED
assert pipeline_run.statuses.count() == 1
# Create another operation run for this pipeline_run
operation_run2 = OperationRunFactory(pipeline_run=pipeline_run)
start_operation_run(operation_run2)
operation_run2.refresh_from_db()
operation_run.set_status(OperationStatuses.SCHEDULED)
pipeline_run.refresh_from_db()
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.SCHEDULED
assert pipeline_run.last_status == PipelineLifeCycle.SCHEDULED
assert pipeline_run.statuses.count() == 2
operation_run.set_status(OperationStatuses.RUNNING)
pipeline_run.refresh_from_db()
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.RUNNING
assert pipeline_run.last_status == PipelineLifeCycle.RUNNING
assert pipeline_run.statuses.count() == 3
operation_run2.set_status(OperationStatuses.SCHEDULED)
assert pipeline_run.last_status == PipelineLifeCycle.RUNNING
assert pipeline_run.statuses.count() == 3
def test_operation_run_creation_sets_created_status(self):
assert JobStatus.objects.count() == 0
# Assert `new_pipeline_run_status` task is also called
operation_run = OperationRunFactory()
start_operation_run(operation_run)
operation_run.refresh_from_db()
assert JobStatus.objects.count() == 1
assert operation_run.last_status == OperationStatuses.CREATED
assert operation_run.entity.last_status == OperationStatuses.CREATED
def test_schedule_start_with_operation_run_already_scheduled_operation_run(
self):
operation_run = OperationRunFactory()
operation_run.status = OperationStatuses.FAILED
operation_run.save()
assert start_operation_run(operation_run) is False
operation_run = OperationRunFactory()
operation_run.status = OperationStatuses.SCHEDULED
operation_run.save()
assert start_operation_run(operation_run) is False
def test_scheduling_operation_run_sets_pipeline_run_to_scheduled(self):
operation_run = OperationRunFactory()
start_operation_run(operation_run)
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.CREATED
pipeline_run = operation_run.pipeline_run
assert pipeline_run.last_status == PipelineLifeCycle.CREATED
assert pipeline_run.statuses.count() == 1
operation_run.set_status(OperationStatuses.SCHEDULED)
operation_run.refresh_from_db()
pipeline_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.SCHEDULED
assert pipeline_run.last_status == PipelineLifeCycle.SCHEDULED
assert pipeline_run.statuses.count() == 2
def test_failed_upstream_operation_runs_sets_pipeline_run_to_finished(
self):
operation_run = OperationRunFactory()
start_operation_run(operation_run)
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.CREATED
pipeline_run = operation_run.pipeline_run
assert pipeline_run.last_status == PipelineLifeCycle.CREATED
assert pipeline_run.statuses.count() == 1
# Stopping the first operation does not stop the pipeline
operation_run.on_upstream_failed()
operation_run.refresh_from_db()
pipeline_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.UPSTREAM_FAILED
assert pipeline_run.last_status == PipelineLifeCycle.DONE
assert pipeline_run.statuses.count() == 2
def test_failed_operation_runs_sets_pipeline_run_to_finished(self):
operation_run = OperationRunFactory()
start_operation_run(operation_run)
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.CREATED
pipeline_run = operation_run.pipeline_run
assert pipeline_run.last_status == PipelineLifeCycle.CREATED
assert pipeline_run.statuses.count() == 1
# Stopping the first operation does not stop the pipeline
operation_run.set_status(OperationStatuses.SCHEDULED)
operation_run.set_status(OperationStatuses.RUNNING)
operation_run.on_failure()
pipeline_run.refresh_from_db()
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.FAILED
assert pipeline_run.last_status == PipelineLifeCycle.DONE
assert pipeline_run.statuses.count() == 4
def test_trigger_policy_one_done(self):
operation_run = OperationRunFactory()
start_operation_run(operation_run)
operation_run.refresh_from_db()
operation = operation_run.operation
operation.trigger_policy = TriggerPolicy.ONE_DONE
operation.save()
# No upstream
assert operation_run.check_upstream_trigger() is False
# Add non done upstream
upstream_run1 = OperationRunFactory()
operation_run.upstream_runs.set([upstream_run1])
assert operation_run.check_upstream_trigger() is False
# A running upstream
upstream_run1.status = OperationStatuses.RUNNING
upstream_run1.save()
assert operation_run.check_upstream_trigger() is False
# A failed upstream
upstream_run1.status = OperationStatuses.FAILED
upstream_run1.save()
assert operation_run.check_upstream_trigger() is True
# Add skipped upstream
upstream_run2 = OperationRunFactory()
operation_run.upstream_runs.set([upstream_run2])
upstream_run2.status = OperationStatuses.SKIPPED
upstream_run2.save()
assert operation_run.check_upstream_trigger() is True
# Add succeeded upstream
upstream_run3 = OperationRunFactory()
operation_run.upstream_runs.set([upstream_run3])
upstream_run3.status = OperationStatuses.SUCCEEDED
upstream_run3.save()
assert operation_run.check_upstream_trigger() is True
# Add another upstream still True
upstream_run4 = OperationRunFactory()
operation_run.upstream_runs.add(upstream_run4)
assert operation_run.check_upstream_trigger() is True
def test_schedule_start_works_with_pipeline_concurrency(self):
operation_run = OperationRunFactory()
operation_run.operation.trigger_policy = TriggerPolicy.ONE_DONE
operation_run.operation.save()
pipeline_run = operation_run.pipeline_run
# Set pipeline concurrency to 1
pipeline_run.pipeline.concurrency = 1
pipeline_run.pipeline.save()
# Add a failed upstream
upstream_run1 = OperationRunFactory(pipeline_run=pipeline_run)
upstream_run2 = OperationRunFactory(pipeline_run=pipeline_run)
assert start_operation_run(upstream_run1) is False
assert start_operation_run(upstream_run2) is True
upstream_run1.refresh_from_db()
upstream_run2.refresh_from_db()
operation_run.upstream_runs.set([upstream_run1, upstream_run2])
with patch('polyflow.tasks.pipelines_start_operation.apply_async'
) as mock_fct:
upstream_run1.set_status(OperationStatuses.FAILED)
assert mock_fct.call_count == 1
operation_run.refresh_from_db()
assert operation_run.last_status is None
upstream_run2.refresh_from_db()
assert upstream_run2.last_status is None # Should be started but e mocked the process
with patch('polyflow.tasks.pipelines_start_operation.apply_async'
) as mock_fct:
assert start_operation_run(upstream_run2) is False
upstream_run2.refresh_from_db()
upstream_run2.set_status(OperationStatuses.RUNNING)
assert mock_fct.call_count == 0
assert start_operation_run(operation_run) is True
assert operation_run.last_status is None
upstream_run2.set_status(OperationStatuses.SUCCEEDED)
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.CREATED
def test_schedule_start_works_with_operation_concurrency(self):
operation_run = OperationRunFactory()
operation_run.operation.trigger_policy = TriggerPolicy.ONE_DONE
operation_run.operation.save()
pipeline_run = operation_run.pipeline_run
# Set operation concurrency to 1
operation_run.operation.concurrency = 1
operation_run.operation.save()
# Add a failed upstream
upstream_run1 = OperationRunFactory(pipeline_run=pipeline_run)
upstream_run2 = OperationRunFactory(pipeline_run=pipeline_run)
assert start_operation_run(upstream_run1) is False
assert start_operation_run(upstream_run2) is False
upstream_run1.refresh_from_db()
upstream_run2.refresh_from_db()
operation_run.upstream_runs.set([upstream_run1, upstream_run2])
with patch('polyflow.tasks.pipelines_start_operation.apply_async'
) as mock_fct:
upstream_run1.set_status(OperationStatuses.FAILED)
upstream_run2.set_status(OperationStatuses.RUNNING)
assert mock_fct.call_count == 1
operation_run.refresh_from_db()
assert operation_run.last_status is None
assert start_operation_run(operation_run) is False
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.CREATED
# Add another operation run for this operation with scheduled
new_operation_run = OperationRunFactory(
operation=operation_run.operation)
new_operation_run.upstream_runs.set([upstream_run1, upstream_run2])
assert new_operation_run.status is None
# Check if we can start another instance
new_operation_run.refresh_from_db()
assert start_operation_run(new_operation_run) is True
new_operation_run.refresh_from_db()
assert new_operation_run.last_status is None
def test_schedule_start_works_when_conditions_are_met_manual(self):
operation_run = OperationRunFactory()
operation_run.operation.trigger_policy = TriggerPolicy.ONE_DONE
operation_run.operation.save()
pipeline_run = operation_run.pipeline_run
# Add a failed upstream
upstream_run1 = OperationRunFactory(pipeline_run=pipeline_run)
assert start_operation_run(upstream_run1) is False
upstream_run1.refresh_from_db()
operation_run.upstream_runs.set([upstream_run1])
with patch('polyflow.tasks.pipelines_start_operation.apply_async'
) as mock_fct:
upstream_run1.set_status(OperationStatuses.FAILED)
assert mock_fct.call_count == 1
operation_run.refresh_from_db()
assert operation_run.last_status is None
assert start_operation_run(operation_run) is False
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.CREATED
def test_schedule_start_with_failed_upstream(self):
operation_run = OperationRunFactory()
operation_run.operation.trigger_policy = TriggerPolicy.ALL_SUCCEEDED
operation_run.operation.save()
# Add a failed upstream
upstream_run1 = OperationRunFactory()
assert start_operation_run(upstream_run1) is False
upstream_run1.refresh_from_db()
operation_run.upstream_runs.set([upstream_run1])
with patch('polyflow.tasks.pipelines_start_operation.apply_async'
) as mock_fct:
upstream_run1.set_status(OperationStatuses.FAILED)
assert mock_fct.call_count == 1
assert start_operation_run(operation_run) is False
# Check also that the task is marked as UPSTREAM_FAILED
# Since this operation cannot be started anymore
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.UPSTREAM_FAILED
def test_schedule_start_works_when_conditions_are_met_auto(self):
operation_run = OperationRunFactory()
operation_run.operation.trigger_policy = TriggerPolicy.ONE_DONE
operation_run.operation.save()
pipeline_run = operation_run.pipeline_run
# Add a failed upstream
upstream_run1 = OperationRunFactory(pipeline_run=pipeline_run)
assert start_operation_run(upstream_run1) is False
upstream_run1.refresh_from_db()
operation_run.upstream_runs.set([upstream_run1])
upstream_run1.set_status(OperationStatuses.FAILED)
operation_run.refresh_from_db()
assert operation_run.last_status == OperationStatuses.CREATED
def test_skipping_pipeline_run_stops_operation_runs(self):
pipeline_run = PipelineRunFactory()
for _ in range(2):
op_run = OperationRunFactory(pipeline_run=pipeline_run)
assert start_operation_run(op_run) is False
assert pipeline_run.statuses.count() == 1
assert pipeline_run.last_status == PipelineLifeCycle.CREATED
assert JobStatus.objects.filter().count() == 2
assert set(JobStatus.objects.values_list('status', flat=True)) == {
OperationStatuses.CREATED,
}
# Set pipeline run to skipped
with patch('scheduler.tasks.jobs.jobs_stop.apply_async'
) as spawner_mock_stop:
pipeline_run.on_skip()
assert pipeline_run.statuses.count() == 2
assert pipeline_run.last_status == PipelineLifeCycle.SKIPPED
# Operation run are also skipped
assert JobStatus.objects.filter().count(
) + spawner_mock_stop.call_count == 6
assert set(JobStatus.objects.values_list('status', flat=True)) == {
OperationStatuses.CREATED, OperationStatuses.SKIPPED
}