def test_check_concurrency(self):
# Pipeline without concurrency defaults to infinite concurrency
pipeline = PipelineFactory()
pipeline_run = PipelineRunFactory(pipeline=pipeline)
assert pipeline_run.check_concurrency() is True
# Pipeline with concurrency and pipeline run with operation runs
pipeline.concurrency = 2
pipeline.save()
# No running operation runs
assert pipeline_run.check_concurrency() is True
# One operation run
operation_run1 = OperationRunFactory(pipeline_run=pipeline_run)
assert pipeline_run.check_concurrency() is True
# One operation run with RUNNING status
operation_run1.status = OperationStatuses.RUNNING
operation_run1.save()
assert pipeline_run.check_concurrency() is True
# Second operation run
operation_run2 = OperationRunFactory(pipeline_run=pipeline_run)
assert pipeline_run.check_concurrency() is True
# Second operation run with RUNNING status
operation_run2.status = OperationStatuses.RUNNING
operation_run2.save()
assert pipeline_run.check_concurrency() is False
def test_pipeline_run_creation_sets_created_status(self):
assert PipelineRunStatus.objects.count() == 0
# Assert `new_pipeline_run_status` task is also called
pipeline_run = PipelineRunFactory()
assert PipelineRunStatus.objects.filter(
pipeline_run=pipeline_run).count() == 1
assert pipeline_run.last_status == PipelineLifeCycle.CREATED
def test_dag_property(self):
pipeline_run = PipelineRunFactory()
operation_runs = [
OperationRunFactory(pipeline_run=pipeline_run) for _ in range(4)
]
operation_runs[0].upstream_runs.set(operation_runs[2:])
operation_runs[1].upstream_runs.set(operation_runs[2:])
operation_by_ids = {op.id: op for op in operation_runs}
assert pipeline_run.dag == ({
operation_runs[0].id: set(),
operation_runs[1].id: set(),
operation_runs[2].id: {operation_runs[0].id, operation_runs[1].id},
operation_runs[3].id: {operation_runs[0].id, operation_runs[1].id},
}, operation_by_ids)
# Add operations outside the dag
operation_run1 = OperationRunFactory()
operation_run1.downstream_runs.set(
[operation_runs[1], operation_runs[2], operation_runs[3]])
operation_run2 = OperationRunFactory()
operation_run2.upstream_runs.set(
[operation_runs[0], operation_runs[2]])
assert pipeline_run.dag == ({
operation_runs[0].id: {
operation_run2.id,
},
operation_runs[1].id: set(),
operation_runs[2].id:
{operation_runs[0].id, operation_runs[1].id, operation_run2.id},
operation_runs[3].id: {operation_runs[0].id, operation_runs[1].id},
}, operation_by_ids)
def setUp(self):
super().setUp()
self.pipeline_run = PipelineRunFactory()
self.tested_events = {
pipeline_run_events.PIPELINE_RUN_CREATED,
pipeline_run_events.PIPELINE_RUN_UPDATED,
pipeline_run_events.PIPELINE_RUN_DELETED,
pipeline_run_events.PIPELINE_RUN_CLEANED_TRIGGERED,
pipeline_run_events.PIPELINE_RUN_VIEWED,
pipeline_run_events.PIPELINE_RUN_ARCHIVED,
pipeline_run_events.PIPELINE_RUN_RESTORED,
pipeline_run_events.PIPELINE_RUN_DELETED_TRIGGERED,
pipeline_run_events.PIPELINE_RUN_STOPPED,
pipeline_run_events.PIPELINE_RUN_STOPPED_TRIGGERED,
pipeline_run_events.PIPELINE_RUN_SKIPPED,
pipeline_run_events.PIPELINE_RUN_SKIPPED_TRIGGERED,
pipeline_run_events.PIPELINE_RUN_RESUMED,
pipeline_run_events.PIPELINE_RUN_RESTARTED,
pipeline_run_events.PIPELINE_RUN_RESTARTED_TRIGGERED,
pipeline_run_events.PIPELINE_RUN_RESUMED_TRIGGERED,
pipeline_run_events.PIPELINE_RUN_NEW_STATUS,
pipeline_run_events.PIPELINE_RUN_SUCCEEDED,
pipeline_run_events.PIPELINE_RUN_FAILED,
pipeline_run_events.PIPELINE_RUN_DONE,
}
def test_stopping_pipeline_run_stops_operation_runs(self):
pipeline_run = PipelineRunFactory()
for _ in range(2):
OperationRunFactory(pipeline_run=pipeline_run)
assert pipeline_run.statuses.count() == 1
assert pipeline_run.last_status == PipelineStatuses.CREATED
assert OperationRunStatus.objects.filter().count() == 2
assert set(OperationRunStatus.objects.values_list(
'status', flat=True)) == {OperationStatuses.CREATED, }
# Set pipeline run to stopped
pipeline_run.on_stop()
assert pipeline_run.statuses.count() == 2
assert pipeline_run.last_status == PipelineStatuses.STOPPED
# Operation run are also stopped
assert OperationRunStatus.objects.filter().count() == 4
assert set(OperationRunStatus.objects.values_list(
'status', flat=True)) == {OperationStatuses.CREATED, OperationStatuses.STOPPED}
def test_stopping_pipeline_run_stops_operation_runs(self):
pipeline_run = PipelineRunFactory()
for _ in range(2):
OperationRunFactory(pipeline_run=pipeline_run)
assert pipeline_run.statuses.count() == 1
assert pipeline_run.last_status == PipelineStatuses.CREATED
assert OperationRunStatus.objects.filter().count() == 2
assert set(OperationRunStatus.objects.values_list(
'status', flat=True)) == {OperationStatuses.CREATED, }
# Set pipeline run to stopped
pipeline_run.on_stop()
assert pipeline_run.statuses.count() == 2
assert pipeline_run.last_status == PipelineStatuses.STOPPED
# Operation run are also stopped
assert OperationRunStatus.objects.filter().count() == 4
assert set(OperationRunStatus.objects.values_list(
'status', flat=True)) == {OperationStatuses.CREATED, OperationStatuses.STOPPED}
def test_check_concurrency(self):
# Pipeline without concurrency defaults to infinite concurrency
pipeline = PipelineFactory()
pipeline_run = PipelineRunFactory(pipeline=pipeline)
assert pipeline_run.check_concurrency() is True
# Pipeline with concurrency and pipeline run with operation runs
pipeline.concurrency = 2
pipeline.save()
# No running operation runs
assert pipeline_run.check_concurrency() is True
# One operation run
operation_run1 = OperationRunFactory(pipeline_run=pipeline_run)
assert pipeline_run.check_concurrency() is True
# One operation run with RUNNING status
OperationRunStatus.objects.create(status=OperationStatuses.RUNNING,
operation_run=operation_run1)
assert pipeline_run.check_concurrency() is True
# Second operation run
operation_run2 = OperationRunFactory(pipeline_run=pipeline_run)
assert pipeline_run.check_concurrency() is True
# Second operation run with RUNNING status
OperationRunStatus.objects.create(status=OperationStatuses.RUNNING,
operation_run=operation_run2)
assert pipeline_run.check_concurrency() is False
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
}