def test_blank_compute_logs(mock_s3_bucket):
with tempfile.TemporaryDirectory() as temp_dir:
manager = S3ComputeLogManager(bucket=mock_s3_bucket.name,
prefix="my_prefix",
local_dir=temp_dir)
# simulate subscription to an in-progress run, where there is no key in the bucket
stdout = manager.read_logs_file("my_run_id", "my_step_key",
ComputeIOType.STDOUT)
stderr = manager.read_logs_file("my_run_id", "my_step_key",
ComputeIOType.STDERR)
assert not stdout.data
assert not stderr.data
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_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(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 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