def ephemeral(tempdir=None):
from dagster.core.launcher.sync_in_memory_run_launcher import SyncInMemoryRunLauncher
from dagster.core.storage.event_log import InMemoryEventLogStorage
from dagster.core.storage.root import LocalArtifactStorage
from dagster.core.storage.runs import InMemoryRunStorage
from dagster.core.storage.noop_compute_log_manager import NoOpComputeLogManager
if tempdir is None:
tempdir = DagsterInstance.temp_storage()
return DagsterInstance(
InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(tempdir),
run_storage=InMemoryRunStorage(),
event_storage=InMemoryEventLogStorage(),
compute_log_manager=NoOpComputeLogManager(),
run_launcher=SyncInMemoryRunLauncher(),
)
def dagster_instance(helm_postgres_url): # pylint: disable=redefined-outer-name
with tempfile.TemporaryDirectory() as tempdir:
with environ({"DAGSTER_HOME": tempdir}):
with DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(tempdir),
run_storage=PostgresRunStorage(helm_postgres_url),
event_storage=PostgresEventLogStorage(helm_postgres_url),
compute_log_manager=NoOpComputeLogManager(),
run_coordinator=DefaultRunCoordinator(),
run_launcher=ExplodingRunLauncher(), # use graphql to launch any runs
ref=InstanceRef.from_dir(tempdir),
) as instance:
yield instance
check_export_runs(instance)
def dagster_instance_for_k8s_run_launcher(
helm_postgres_url_for_k8s_run_launcher, run_launcher): # pylint: disable=redefined-outer-name
tempdir = DagsterInstance.temp_storage()
with DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(tempdir),
run_storage=PostgresRunStorage(
helm_postgres_url_for_k8s_run_launcher),
event_storage=PostgresEventLogStorage(
helm_postgres_url_for_k8s_run_launcher),
schedule_storage=PostgresScheduleStorage(
helm_postgres_url_for_k8s_run_launcher),
compute_log_manager=NoOpComputeLogManager(),
run_coordinator=DefaultRunCoordinator(),
run_launcher=run_launcher,
) as instance:
yield instance
def dagster_instance(helm_namespace, run_launcher): # pylint: disable=redefined-outer-name
tempdir = DagsterInstance.temp_storage()
with local_port_forward_postgres(
namespace=helm_namespace) as local_forward_port:
postgres_url = 'postgresql://*****:*****@localhost:{local_forward_port}/test'.format(
local_forward_port=local_forward_port)
print('Local Postgres forwarding URL: ', postgres_url)
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(tempdir),
run_storage=PostgresRunStorage(postgres_url),
event_storage=PostgresEventLogStorage(postgres_url),
compute_log_manager=NoOpComputeLogManager(),
run_launcher=run_launcher,
)
yield instance
def dagster_instance_with_k8s_scheduler(helm_postgres_url_for_k8s_run_launcher,
run_launcher, k8s_scheduler,
schedule_tempdir): # pylint: disable=redefined-outer-name
with DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(schedule_tempdir),
run_storage=SqliteRunStorage.from_local(
os.path.join(schedule_tempdir, "runs")),
event_storage=PostgresEventLogStorage(
helm_postgres_url_for_k8s_run_launcher),
compute_log_manager=NoOpComputeLogManager(),
run_coordinator=DefaultRunCoordinator(),
run_launcher=run_launcher,
schedule_storage=SqliteScheduleStorage.from_local(
os.path.join(schedule_tempdir, "schedules")),
scheduler=k8s_scheduler,
) as instance:
yield instance
def _postgres_instance_with_grpc_api_hijack():
with seven.TemporaryDirectory() as temp_dir:
with graphql_postgres_instance() as pg_conn_string:
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=TestPostgresInstance.clean_run_storage(pg_conn_string),
event_storage=TestPostgresInstance.clean_event_log_storage(pg_conn_string),
compute_log_manager=LocalComputeLogManager(temp_dir),
run_launcher=EphemeralGrpcRunLauncher(),
schedule_storage=TestPostgresInstance.clean_schedule_storage(
pg_conn_string
),
)
try:
yield instance
finally:
instance.run_launcher.join()
def test_basic_start_scheduled_execution_with_run_launcher():
test_queue = InMemoryRunLauncher()
with seven.TemporaryDirectory() as temp_dir:
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=InMemoryRunStorage(),
event_storage=InMemoryEventLogStorage(),
compute_log_manager=NoOpComputeLogManager(temp_dir),
run_launcher=test_queue,
)
context = define_context_for_repository_yaml(path=file_relative_path(
__file__, '../repository.yaml'),
instance=instance)
scheduler_handle = context.scheduler_handle
scheduler_handle.up(python_path=sys.executable,
repository_path=file_relative_path(
__file__, '../'))
result = execute_dagster_graphql(
context,
START_SCHEDULED_EXECUTION_QUERY,
variables={'scheduleName': 'no_config_pipeline_hourly_schedule'},
)
assert not result.errors
assert result.data
# just test existence
assert (result.data['startScheduledExecution']['__typename'] ==
'LaunchPipelineExecutionSuccess')
assert uuid.UUID(
result.data['startScheduledExecution']['run']['runId'])
assert (result.data['startScheduledExecution']['run']['pipeline']
['name'] == 'no_config_pipeline')
assert any(
tag['key'] == 'dagster/schedule_name'
and tag['value'] == 'no_config_pipeline_hourly_schedule'
for tag in result.data['startScheduledExecution']['run']['tags'])
def test_run_launcher():
test_queue = InMemoryRunLauncher()
with seven.TemporaryDirectory() as temp_dir:
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=InMemoryRunStorage(),
event_storage=InMemoryEventLogStorage(),
compute_log_manager=NoOpComputeLogManager(temp_dir),
run_launcher=test_queue,
)
context = define_context_for_repository_yaml(path=file_relative_path(
__file__, '../repository.yaml'),
instance=instance)
result = execute_dagster_graphql(
context=context,
query=LAUNCH_PIPELINE_EXECUTION_MUTATION,
variables={
'executionParams': {
'selector': {
'name': 'no_config_pipeline'
},
'mode': 'default'
}
},
)
assert result.data['launchPipelineExecution'][
'__typename'] == 'LaunchPipelineRunSuccess'
assert result.data['launchPipelineExecution']['run'][
'status'] == 'NOT_STARTED'
run_id = result.data['launchPipelineExecution']['run']['runId']
test_queue.run_one(instance)
result = execute_dagster_graphql(context=context,
query=RUN_QUERY,
variables={'runId': run_id})
assert result.data['pipelineRunOrError']['__typename'] == 'PipelineRun'
assert result.data['pipelineRunOrError']['status'] == 'SUCCESS'
def test_start_stop_schedule():
with seven.TemporaryDirectory() as temp_dir:
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=InMemoryRunStorage(),
event_storage=InMemoryEventLogStorage(),
compute_log_manager=NoOpComputeLogManager(temp_dir),
schedule_storage=SqliteScheduleStorage.from_local(temp_dir),
scheduler=FilesystemTestScheduler(temp_dir),
run_launcher=SyncInMemoryRunLauncher(),
)
context = define_context_for_repository_yaml(path=file_relative_path(
__file__, '../repository.yaml'),
instance=instance)
# Initialize scheduler
repository = context.legacy_get_repository_definition()
reconcile_scheduler_state(
python_path=sys.executable,
repository_path="",
repository=repository,
instance=instance,
)
# Start schedule
start_result = execute_dagster_graphql(
context,
START_SCHEDULES_QUERY,
variables={'scheduleName': 'no_config_pipeline_hourly_schedule'},
)
assert start_result.data['startSchedule']['schedule'][
'status'] == 'RUNNING'
# Stop schedule
stop_result = execute_dagster_graphql(
context,
STOP_SCHEDULES_QUERY,
variables={'scheduleName': 'no_config_pipeline_hourly_schedule'},
)
assert stop_result.data['stopRunningSchedule']['schedule'][
'status'] == 'STOPPED'
def test_get_all_schedules():
with seven.TemporaryDirectory() as temp_dir:
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=InMemoryRunStorage(),
event_storage=InMemoryEventLogStorage(),
compute_log_manager=NoOpComputeLogManager(temp_dir),
schedule_storage=SqliteScheduleStorage.from_local(temp_dir),
scheduler=FilesytemTestScheduler(temp_dir),
)
context = define_context_for_repository_yaml(path=file_relative_path(
__file__, '../repository.yaml'),
instance=instance)
# Initialize scheduler
repository = context.get_repository()
scheduler_handle = context.scheduler_handle
scheduler_handle.up(
python_path=sys.executable,
repository_path="",
repository=repository,
instance=instance,
)
# Start schedule
schedule = instance.start_schedule(
repository, "no_config_pipeline_hourly_schedule")
# Query Scheduler + all Schedules
scheduler_result = execute_dagster_graphql(context,
GET_SCHEDULES_QUERY)
assert scheduler_result.data
assert scheduler_result.data['scheduler']
assert scheduler_result.data['scheduler']['runningSchedules']
assert len(
scheduler_result.data['scheduler']['runningSchedules']) == 11
for schedule in scheduler_result.data['scheduler']['runningSchedules']:
assert (schedule['scheduleDefinition']['environmentConfigYaml'] ==
'storage:\n filesystem: {}\n')
def dagster_instance_for_user_deployments_subchart_disabled(
helm_postgres_url_for_user_deployments_subchart_disabled,
): # pylint: disable=redefined-outer-name
tempdir = DagsterInstance.temp_storage()
with DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(tempdir),
run_storage=PostgresRunStorage(helm_postgres_url_for_user_deployments_subchart_disabled),
event_storage=PostgresEventLogStorage(
helm_postgres_url_for_user_deployments_subchart_disabled
),
compute_log_manager=NoOpComputeLogManager(),
run_coordinator=DefaultRunCoordinator(),
run_launcher=ExplodingRunLauncher(),
) as instance:
yield instance
check_export_runs(instance)
def dagster_instance_for_daemon(
helm_postgres_url_for_daemon,
): # pylint: disable=redefined-outer-name
tempdir = DagsterInstance.temp_storage()
with DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(tempdir),
run_storage=PostgresRunStorage(helm_postgres_url_for_daemon),
event_storage=PostgresEventLogStorage(helm_postgres_url_for_daemon),
schedule_storage=PostgresScheduleStorage(helm_postgres_url_for_daemon),
compute_log_manager=NoOpComputeLogManager(),
run_coordinator=QueuedRunCoordinator(),
run_launcher=ExplodingRunLauncher(),
scheduler=DagsterDaemonScheduler(),
) as instance:
yield instance
check_export_runs(instance)
def run_launch(execution_args, expected_count=None):
runner = CliRunner()
run_launcher = InMemoryRunLauncher()
with seven.TemporaryDirectory() as temp_dir:
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=InMemoryRunStorage(),
event_storage=InMemoryEventLogStorage(),
compute_log_manager=NoOpComputeLogManager(),
run_launcher=run_launcher,
)
with mock.patch('dagster.core.instance.DagsterInstance.get') as _instance:
_instance.return_value = instance
result = runner.invoke(pipeline_launch_command, execution_args)
assert result.exit_code == 0, result.stdout
if expected_count:
assert len(run_launcher.queue()) == expected_count
def test_fs_stores():
@pipeline
def simple():
@solid
def easy(context):
context.log.info("easy")
return "easy"
easy()
with tempfile.TemporaryDirectory() as temp_dir:
with environ({"DAGSTER_HOME": temp_dir}):
run_store = SqliteRunStorage.from_local(temp_dir)
event_store = SqliteEventLogStorage(temp_dir)
compute_log_manager = LocalComputeLogManager(temp_dir)
instance = DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=run_store,
event_storage=event_store,
compute_log_manager=compute_log_manager,
run_coordinator=DefaultRunCoordinator(),
run_launcher=DefaultRunLauncher(),
ref=InstanceRef.from_dir(temp_dir),
settings={"telemetry": {
"enabled": False
}},
)
result = execute_pipeline(simple, instance=instance)
assert run_store.has_run(result.run_id)
assert run_store.get_run_by_id(
result.run_id).status == PipelineRunStatus.SUCCESS
assert DagsterEventType.PIPELINE_SUCCESS in [
event.dagster_event.event_type
for event in event_store.get_logs_for_run(result.run_id)
if event.is_dagster_event
]
stats = event_store.get_stats_for_run(result.run_id)
assert stats.steps_succeeded == 1
assert stats.end_time is not None
def ephemeral(tempdir=None):
from dagster.core.storage.event_log import InMemoryEventLogStorage
from dagster.core.storage.root import LocalArtifactStorage
from dagster.core.storage.runs import InMemoryRunStorage
from dagster.core.storage.local_compute_log_manager import NoOpComputeLogManager
if tempdir is None:
tempdir = DagsterInstance.temp_storage()
feature_set = dagster_feature_set(tempdir)
return DagsterInstance(
InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(tempdir),
run_storage=InMemoryRunStorage(),
event_storage=InMemoryEventLogStorage(),
compute_log_manager=NoOpComputeLogManager(
_compute_logs_directory(tempdir)),
feature_set=feature_set,
)
def dagster_instance_for_k8s_run_launcher(
helm_postgres_url_for_k8s_run_launcher,
): # pylint: disable=redefined-outer-name
tempdir = DagsterInstance.temp_storage()
instance_ref = InstanceRef.from_dir(tempdir)
with DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(tempdir),
run_storage=PostgresRunStorage(helm_postgres_url_for_k8s_run_launcher),
event_storage=PostgresEventLogStorage(helm_postgres_url_for_k8s_run_launcher),
schedule_storage=PostgresScheduleStorage(helm_postgres_url_for_k8s_run_launcher),
compute_log_manager=NoOpComputeLogManager(),
run_coordinator=DefaultRunCoordinator(),
run_launcher=ExplodingRunLauncher(),
ref=instance_ref,
) as instance:
yield instance
check_export_runs(instance)
def dagster_instance_with_k8s_scheduler(helm_namespace, run_launcher,
k8s_scheduler, schedule_tempdir):
with local_port_forward_postgres(
namespace=helm_namespace) as local_forward_port:
postgres_url = 'postgresql://*****:*****@localhost:{local_forward_port}/test'.format(
local_forward_port=local_forward_port)
print('Local Postgres forwarding URL: ', postgres_url)
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(schedule_tempdir),
run_storage=SqliteRunStorage.from_local(
os.path.join(schedule_tempdir, 'runs')),
event_storage=PostgresEventLogStorage(postgres_url),
compute_log_manager=NoOpComputeLogManager(),
run_launcher=run_launcher,
schedule_storage=SqliteScheduleStorage.from_local(
os.path.join(schedule_tempdir, 'schedules')),
scheduler=k8s_scheduler,
)
yield instance
def dagster_instance_for_k8s_run_launcher(helm_namespace_for_k8s_run_launcher, run_launcher):
tempdir = DagsterInstance.temp_storage()
with local_port_forward_postgres(
namespace=helm_namespace_for_k8s_run_launcher
) as local_forward_port:
postgres_url = "postgresql://*****:*****@localhost:{local_forward_port}/test".format(
local_forward_port=local_forward_port
)
print("Local Postgres forwarding URL: ", postgres_url)
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(tempdir),
run_storage=PostgresRunStorage(postgres_url),
event_storage=PostgresEventLogStorage(postgres_url),
schedule_storage=PostgresScheduleStorage(postgres_url),
compute_log_manager=NoOpComputeLogManager(),
run_launcher=run_launcher,
)
yield instance
def dagster_instance_for_daemon(helm_namespace_for_daemon, run_launcher): # pylint: disable=redefined-outer-name
tempdir = DagsterInstance.temp_storage()
with local_port_forward_postgres(
namespace=helm_namespace_for_daemon) as local_forward_port:
postgres_url = "postgresql://*****:*****@localhost:{local_forward_port}/test".format(
local_forward_port=local_forward_port)
print("Local Postgres forwarding URL: ", postgres_url)
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(tempdir),
run_storage=PostgresRunStorage(postgres_url),
event_storage=PostgresEventLogStorage(postgres_url),
schedule_storage=PostgresScheduleStorage(postgres_url),
compute_log_manager=NoOpComputeLogManager(),
run_coordinator=QueuedRunCoordinator(),
run_launcher=run_launcher,
scheduler=DagsterDaemonScheduler(),
)
yield instance
def test_run_record_timestamps(self, storage):
assert storage
self._skip_in_memory(storage)
@op
def a():
pass
@job
def my_job():
a()
with tempfile.TemporaryDirectory() as temp_dir:
if storage._instance: # pylint: disable=protected-access
instance = storage._instance # pylint: disable=protected-access
else:
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=storage,
event_storage=InMemoryEventLogStorage(),
compute_log_manager=NoOpComputeLogManager(),
run_coordinator=DefaultRunCoordinator(),
run_launcher=SyncInMemoryRunLauncher(),
)
freeze_datetime = to_timezone(
create_pendulum_time(2019, 11, 2, 0, 0, 0, tz="US/Central"), "US/Pacific"
)
with pendulum.test(freeze_datetime):
result = my_job.execute_in_process(instance=instance)
records = instance.get_run_records(
filters=PipelineRunsFilter(run_ids=[result.run_id])
)
assert len(records) == 1
record = records[0]
assert record.start_time == freeze_datetime.timestamp()
assert record.end_time == freeze_datetime.timestamp()
def test_compute_log_manager_skip_empty_upload(mock_s3_bucket):
@op
def easy(context):
context.log.info("easy")
@job
def simple():
easy()
with tempfile.TemporaryDirectory() as temp_dir:
with environ({"DAGSTER_HOME": temp_dir}):
run_store = SqliteRunStorage.from_local(temp_dir)
event_store = SqliteEventLogStorage(temp_dir)
PREFIX = "my_prefix"
manager = S3ComputeLogManager(bucket=mock_s3_bucket.name,
prefix=PREFIX,
skip_empty_files=True)
instance = DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=run_store,
event_storage=event_store,
compute_log_manager=manager,
run_coordinator=DefaultRunCoordinator(),
run_launcher=DefaultRunLauncher(),
ref=InstanceRef.from_dir(temp_dir),
)
result = simple.execute_in_process(instance=instance)
stderr_object = mock_s3_bucket.Object(
key=f"{PREFIX}/storage/{result.run_id}/compute_logs/easy.err"
).get()
assert stderr_object
with pytest.raises(ClientError):
# stdout is not uploaded because we do not print anything to stdout
mock_s3_bucket.Object(
key=f"{PREFIX}/storage/{result.run_id}/compute_logs/easy.out"
).get()
def test_fs_stores():
@pipeline
def simple():
@solid
def easy(context):
context.log.info('easy')
return 'easy'
easy()
with seven.TemporaryDirectory() as temp_dir:
run_store = SqliteRunStorage.from_local(temp_dir)
event_store = SqliteEventLogStorage(temp_dir)
compute_log_manager = LocalComputeLogManager(temp_dir)
instance = DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=run_store,
event_storage=event_store,
compute_log_manager=compute_log_manager,
)
run = RunConfig()
execute_pipeline(simple, run_config=run, instance=instance)
assert run_store.has_run(run.run_id)
assert run_store.get_run_by_id(
run.run_id).status == PipelineRunStatus.SUCCESS
assert DagsterEventType.PIPELINE_SUCCESS in [
event.dagster_event.event_type
for event in event_store.get_logs_for_run(run.run_id)
if event.is_dagster_event
]
stats = event_store.get_stats_for_run(run.run_id)
assert stats.steps_succeeded == 1
assert stats.end_time is not None
def test_compute_log_manager(s3_bucket):
@pipeline
def simple():
@solid
def easy(context):
context.log.info('easy')
print(HELLO_WORLD)
return 'easy'
easy()
# Uses mock S3
s3 = boto3.client('s3')
s3.create_bucket(Bucket=s3_bucket)
with seven.TemporaryDirectory() as temp_dir:
run_store = SqliteRunStorage.from_local(temp_dir)
event_store = SqliteEventLogStorage(temp_dir)
manager = S3ComputeLogManager(bucket=s3_bucket,
prefix='my_prefix',
local_dir=temp_dir)
instance = DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=run_store,
event_storage=event_store,
compute_log_manager=manager,
run_launcher=CliApiRunLauncher(),
)
result = execute_pipeline(simple, instance=instance)
compute_steps = [
event.step_key for event in result.step_event_list
if event.event_type == DagsterEventType.STEP_START
]
assert len(compute_steps) == 1
step_key = compute_steps[0]
stdout = manager.read_logs_file(result.run_id, step_key,
ComputeIOType.STDOUT)
assert stdout.data == HELLO_WORLD + SEPARATOR
stderr = manager.read_logs_file(result.run_id, step_key,
ComputeIOType.STDERR)
for expected in EXPECTED_LOGS:
assert expected in stderr.data
# Check S3 directly
s3_object = s3.get_object(
Bucket=s3_bucket,
Key='{prefix}/storage/{run_id}/compute_logs/easy.compute.err'.
format(prefix='my_prefix', run_id=result.run_id),
)
stderr_s3 = six.ensure_str(s3_object['Body'].read())
for expected in EXPECTED_LOGS:
assert expected in stderr_s3
# Check download behavior by deleting locally cached logs
compute_logs_dir = os.path.join(temp_dir, result.run_id,
'compute_logs')
for filename in os.listdir(compute_logs_dir):
os.unlink(os.path.join(compute_logs_dir, filename))
stdout = manager.read_logs_file(result.run_id, step_key,
ComputeIOType.STDOUT)
assert stdout.data == HELLO_WORLD + SEPARATOR
stderr = manager.read_logs_file(result.run_id, step_key,
ComputeIOType.STDERR)
for expected in EXPECTED_LOGS:
assert expected in stderr.data
def dagster_instance(helm_namespace, run_launcher): # pylint: disable=redefined-outer-name
@contextmanager
def local_port_forward_postgres():
print('Port-forwarding postgres')
postgres_pod_name = (check_output([
'kubectl',
'get',
'pods',
'--namespace',
helm_namespace,
'-l',
'app=postgresql,release=dagster',
'-o',
'jsonpath="{.items[0].metadata.name}"',
]).decode('utf-8').strip('"'))
forward_port = find_free_port()
wait_for_pod(postgres_pod_name, namespace=helm_namespace)
try:
p = subprocess.Popen([
'kubectl',
'port-forward',
'--namespace',
helm_namespace,
postgres_pod_name,
'{forward_port}:5432'.format(forward_port=forward_port),
])
# Validate port forwarding works
start = time.time()
while True:
if time.time() - start > PG_PORT_FORWARDING_TIMEOUT:
raise Exception(
'Timed out while waiting for postgres port forwarding')
print(
'Waiting for port forwarding from k8s pod %s:5432 to localhost:%d to be'
' available...' % (postgres_pod_name, forward_port))
try:
conn = psycopg2.connect(
database='test',
user='******',
password='******',
host='localhost',
port=forward_port,
)
conn.close()
break
except: # pylint: disable=bare-except, broad-except
time.sleep(1)
continue
yield forward_port
finally:
print('Terminating port-forwarding')
p.terminate()
tempdir = DagsterInstance.temp_storage()
with local_port_forward_postgres() as local_forward_port:
postgres_url = 'postgresql://*****:*****@localhost:{local_forward_port}/test'.format(
local_forward_port=local_forward_port)
print('Local Postgres forwarding URL: ', postgres_url)
instance = DagsterInstance(
instance_type=InstanceType.EPHEMERAL,
local_artifact_storage=LocalArtifactStorage(tempdir),
run_storage=PostgresRunStorage(postgres_url),
event_storage=PostgresEventLogStorage(postgres_url),
compute_log_manager=NoOpComputeLogManager(),
run_launcher=run_launcher,
)
yield instance
def test_compute_log_manager_with_envvar(gcs_bucket):
@job
def simple():
@op
def easy(context):
context.log.info("easy")
print(HELLO_WORLD) # pylint: disable=print-call
return "easy"
easy()
with open(os.environ.get("GOOGLE_APPLICATION_CREDENTIALS")) as f:
with tempfile.TemporaryDirectory() as temp_dir:
with environ({"ENV_VAR": f.read(), "DAGSTER_HOME": temp_dir}):
run_store = SqliteRunStorage.from_local(temp_dir)
event_store = SqliteEventLogStorage(temp_dir)
manager = GCSComputeLogManager(
bucket=gcs_bucket,
prefix="my_prefix",
local_dir=temp_dir,
json_credentials_envvar="ENV_VAR",
)
instance = DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=run_store,
event_storage=event_store,
compute_log_manager=manager,
run_coordinator=DefaultRunCoordinator(),
run_launcher=DefaultRunLauncher(),
ref=InstanceRef.from_dir(temp_dir),
)
result = simple.execute_in_process(instance=instance)
compute_steps = [
event.step_key
for event in result.all_node_events
if event.event_type == DagsterEventType.STEP_START
]
assert len(compute_steps) == 1
step_key = compute_steps[0]
stdout = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDOUT)
assert stdout.data == HELLO_WORLD + SEPARATOR
stderr = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDERR)
for expected in EXPECTED_LOGS:
assert expected in stderr.data
# Check GCS directly
stderr_gcs = (
storage.Client()
.bucket(gcs_bucket)
.blob(f"my_prefix/storage/{result.run_id}/compute_logs/easy.err")
.download_as_bytes()
.decode("utf-8")
)
for expected in EXPECTED_LOGS:
assert expected in stderr_gcs
# Check download behavior by deleting locally cached logs
compute_logs_dir = os.path.join(temp_dir, result.run_id, "compute_logs")
for filename in os.listdir(compute_logs_dir):
os.unlink(os.path.join(compute_logs_dir, filename))
stdout = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDOUT)
assert stdout.data == HELLO_WORLD + SEPARATOR
stderr = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDERR)
for expected in EXPECTED_LOGS:
assert expected in stderr.data
def test_dev_loop_changing_versions():
with seven.TemporaryDirectory() as temp_dir:
run_store = SqliteRunStorage.from_local(temp_dir)
event_store = ConsolidatedSqliteEventLogStorage(temp_dir)
compute_log_manager = LocalComputeLogManager(temp_dir)
instance = DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=run_store,
event_storage=event_store,
compute_log_manager=compute_log_manager,
run_launcher=CliApiRunLauncher(),
)
run_config = {
"solids": {
"create_string_1": {"config": {"input_str": "apple", "base_dir": temp_dir}},
"create_string_2": {"config": {"input_str": "apple", "base_dir": temp_dir}},
"take_string_1": {"config": {"input_str": "apple", "base_dir": temp_dir}},
"take_string_2": {"config": {"input_str": "apple", "base_dir": temp_dir}},
"take_string_two_inputs": {"config": {"input_str": "apple", "base_dir": temp_dir}},
},
"intermediate_storage": {"filesystem": {"config": {"base_dir": temp_dir}}},
}
result = execute_pipeline(
basic_pipeline,
run_config=run_config,
mode="only_mode",
tags={"dagster/is_memoized_run": "true"},
instance=instance,
)
assert result.success
assert not get_step_keys_to_execute(instance, basic_pipeline, run_config, "only_mode")
run_config["solids"]["take_string_1"]["config"]["input_str"] = "banana"
assert set(
get_step_keys_to_execute(instance, basic_pipeline, run_config, "only_mode")
) == set(["take_string_1.compute", "take_string_two_inputs.compute"])
result2 = execute_pipeline(
basic_pipeline,
run_config=run_config,
mode="only_mode",
tags={"dagster/is_memoized_run": "true"},
instance=instance,
)
assert result2.success
assert not get_step_keys_to_execute(instance, basic_pipeline, run_config, "only_mode")
run_config["solids"]["take_string_two_inputs"]["config"]["input_str"] = "banana"
assert get_step_keys_to_execute(instance, basic_pipeline, run_config, "only_mode") == [
"take_string_two_inputs.compute"
]
result3 = execute_pipeline(
basic_pipeline,
run_config=run_config,
mode="only_mode",
tags={"dagster/is_memoized_run": "true"},
instance=instance,
)
assert result3.success
assert not get_step_keys_to_execute(instance, basic_pipeline, run_config, "only_mode")
def test_execute_display_command():
with tempfile.TemporaryDirectory() as temp_dir:
run_store = SqliteRunStorage.from_local(temp_dir)
event_store = ConsolidatedSqliteEventLogStorage(temp_dir)
compute_log_manager = LocalComputeLogManager(temp_dir)
instance = DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=run_store,
event_storage=event_store,
compute_log_manager=compute_log_manager,
run_coordinator=DefaultRunCoordinator(),
run_launcher=DefaultRunLauncher(),
)
run_config = {
"solids": {
"create_string_1_asset": {
"config": {
"input_str": "apple"
}
},
"take_string_1_asset": {
"config": {
"input_str": "apple"
}
},
},
"resources": {
"io_manager": {
"config": {
"base_dir": temp_dir
}
}
},
}
# write run config to temp file
# file is temp because intermediate storage directory is temporary
with open(os.path.join(temp_dir, "pipeline_config.yaml"), "w") as f:
f.write(yaml.dump(run_config))
kwargs = {
"config": (os.path.join(temp_dir, "pipeline_config.yaml"), ),
"pipeline":
"asset_pipeline",
"python_file":
file_relative_path(
__file__,
"../../execution_tests/memoized_dev_loop_pipeline.py"),
"tags":
'{"dagster/is_memoized_run": "true"}',
}
with Capturing() as output:
execute_list_versions_command(kwargs=kwargs, instance=instance)
assert output
# execute the pipeline once so that addresses have been populated.
result = execute_pipeline(
asset_pipeline,
run_config=run_config,
mode="only_mode",
tags={"dagster/is_memoized_run": "true"},
instance=instance,
)
assert result.success
with Capturing() as output:
execute_list_versions_command(kwargs=kwargs, instance=instance)
assert output
def test_compute_log_manager(mock_s3_bucket):
@op
def easy(context):
context.log.info("easy")
print(HELLO_WORLD) # pylint: disable=print-call
return "easy"
@job
def simple():
easy()
with tempfile.TemporaryDirectory() as temp_dir:
with environ({"DAGSTER_HOME": temp_dir}):
run_store = SqliteRunStorage.from_local(temp_dir)
event_store = SqliteEventLogStorage(temp_dir)
manager = S3ComputeLogManager(bucket=mock_s3_bucket.name,
prefix="my_prefix",
local_dir=temp_dir)
instance = DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=run_store,
event_storage=event_store,
compute_log_manager=manager,
run_coordinator=DefaultRunCoordinator(),
run_launcher=DefaultRunLauncher(),
ref=InstanceRef.from_dir(temp_dir),
)
result = simple.execute_in_process(instance=instance)
compute_steps = [
event.step_key for event in result.all_node_events
if event.event_type == DagsterEventType.STEP_START
]
assert len(compute_steps) == 1
step_key = compute_steps[0]
stdout = manager.read_logs_file(result.run_id, step_key,
ComputeIOType.STDOUT)
assert stdout.data == HELLO_WORLD + SEPARATOR
stderr = manager.read_logs_file(result.run_id, step_key,
ComputeIOType.STDERR)
for expected in EXPECTED_LOGS:
assert expected in stderr.data
# Check S3 directly
s3_object = mock_s3_bucket.Object(
key=f"my_prefix/storage/{result.run_id}/compute_logs/easy.err")
stderr_s3 = s3_object.get()["Body"].read().decode("utf-8")
for expected in EXPECTED_LOGS:
assert expected in stderr_s3
# Check download behavior by deleting locally cached logs
compute_logs_dir = os.path.join(temp_dir, result.run_id,
"compute_logs")
for filename in os.listdir(compute_logs_dir):
os.unlink(os.path.join(compute_logs_dir, filename))
stdout = manager.read_logs_file(result.run_id, step_key,
ComputeIOType.STDOUT)
assert stdout.data == HELLO_WORLD + SEPARATOR
stderr = manager.read_logs_file(result.run_id, step_key,
ComputeIOType.STDERR)
for expected in EXPECTED_LOGS:
assert expected in stderr.data
def test_compute_log_manager(mock_s3_bucket):
@pipeline
def simple():
@solid
def easy(context):
context.log.info("easy")
print(HELLO_WORLD) # pylint: disable=print-call
return "easy"
easy()
with seven.TemporaryDirectory() as temp_dir:
run_store = SqliteRunStorage.from_local(temp_dir)
event_store = SqliteEventLogStorage(temp_dir)
manager = S3ComputeLogManager(
bucket=mock_s3_bucket.name, prefix="my_prefix", local_dir=temp_dir
)
instance = DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=run_store,
event_storage=event_store,
compute_log_manager=manager,
run_coordinator=DefaultRunCoordinator(),
run_launcher=DefaultRunLauncher(),
)
result = execute_pipeline(simple, instance=instance)
compute_steps = [
event.step_key
for event in result.step_event_list
if event.event_type == DagsterEventType.STEP_START
]
assert len(compute_steps) == 1
step_key = compute_steps[0]
stdout = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDOUT)
assert stdout.data == HELLO_WORLD + SEPARATOR
stderr = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDERR)
for expected in EXPECTED_LOGS:
assert expected in stderr.data
# Check S3 directly
s3_object = mock_s3_bucket.Object(
key="{prefix}/storage/{run_id}/compute_logs/easy.err".format(
prefix="my_prefix", run_id=result.run_id
),
)
stderr_s3 = six.ensure_str(s3_object.get()["Body"].read())
for expected in EXPECTED_LOGS:
assert expected in stderr_s3
# Check download behavior by deleting locally cached logs
compute_logs_dir = os.path.join(temp_dir, result.run_id, "compute_logs")
for filename in os.listdir(compute_logs_dir):
os.unlink(os.path.join(compute_logs_dir, filename))
stdout = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDOUT)
assert stdout.data == HELLO_WORLD + SEPARATOR
stderr = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDERR)
for expected in EXPECTED_LOGS:
assert expected in stderr.data
def test_compute_log_manager(
mock_create_blob_client, mock_generate_blob_sas, storage_account, container, credential
):
mock_generate_blob_sas.return_value = "fake-url"
fake_client = FakeBlobServiceClient(storage_account)
mock_create_blob_client.return_value = fake_client
@pipeline
def simple():
@solid
def easy(context):
context.log.info("easy")
print(HELLO_WORLD) # pylint: disable=print-call
return "easy"
easy()
with tempfile.TemporaryDirectory() as temp_dir:
run_store = SqliteRunStorage.from_local(temp_dir)
event_store = SqliteEventLogStorage(temp_dir)
manager = AzureBlobComputeLogManager(
storage_account=storage_account,
container=container,
prefix="my_prefix",
local_dir=temp_dir,
secret_key=credential,
)
instance = DagsterInstance(
instance_type=InstanceType.PERSISTENT,
local_artifact_storage=LocalArtifactStorage(temp_dir),
run_storage=run_store,
event_storage=event_store,
compute_log_manager=manager,
run_coordinator=DefaultRunCoordinator(),
run_launcher=SyncInMemoryRunLauncher(),
)
result = execute_pipeline(simple, instance=instance)
compute_steps = [
event.step_key
for event in result.step_event_list
if event.event_type == DagsterEventType.STEP_START
]
assert len(compute_steps) == 1
step_key = compute_steps[0]
stdout = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDOUT)
assert stdout.data == HELLO_WORLD + SEPARATOR
stderr = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDERR)
for expected in EXPECTED_LOGS:
assert expected in stderr.data
# Check ADLS2 directly
adls2_object = fake_client.get_blob_client(
container=container,
blob="{prefix}/storage/{run_id}/compute_logs/easy.err".format(
prefix="my_prefix", run_id=result.run_id
),
)
adls2_stderr = adls2_object.download_blob().readall().decode("utf-8")
for expected in EXPECTED_LOGS:
assert expected in adls2_stderr
# Check download behavior by deleting locally cached logs
compute_logs_dir = os.path.join(temp_dir, result.run_id, "compute_logs")
for filename in os.listdir(compute_logs_dir):
os.unlink(os.path.join(compute_logs_dir, filename))
stdout = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDOUT)
assert stdout.data == HELLO_WORLD + SEPARATOR
stderr = manager.read_logs_file(result.run_id, step_key, ComputeIOType.STDERR)
for expected in EXPECTED_LOGS:
assert expected in stderr.data
