def test_run_helloworld_sync(sync_service, specfile, state):
"""Execute the helloworld example."""
# -- Setup ----------------------------------------------------------------
#
# Start a new run for the workflow template.
with sync_service() as api:
workflow_id = create_workflow(api,
source=TEMPLATE_DIR,
specfile=specfile)
user_id = create_user(api)
with sync_service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
names = io_file(data=['Alice', 'Bob'], format='plain/text')
file_id = upload_file(api, group_id, names)
args = [
serialize_arg('names', serialize_fh(file_id, 'data/names.txt')),
serialize_arg('sleeptime', 3)
]
run_id = start_run(api, group_id, arguments=args)
# -- Validate the run handle against the expected state -------------------
with sync_service(user_id=user_id) as api:
r = api.runs().get_run(run_id)
serialize.validate_run_handle(r, state=state)
if state == st.STATE_SUCCESS:
# The run should have the greetings.txt file as a result.
files = dict()
for obj in r['files']:
files[obj['name']] = obj['id']
assert len(files) == 1
fh = api.runs().get_result_file(
run_id=run_id, file_id=files['results/greetings.txt'])
value = fh.open().read().decode('utf-8').strip()
assert 'Hello Alice!' in value
assert 'Hello Bob!' in value
def test_run_remote_workflow_with_error(tmpdir):
"""Execute the remote workflow example that will end in an error state in
asynchronous mode.
"""
# -- Setup ----------------------------------------------------------------
#
# Start a new run for the workflow template.
env = Config().basedir(tmpdir)
engine = RemoteTestController(client=RemoteTestClient(runcount=3,
error='some error'),
poll_interval=1,
is_async=True)
service = LocalAPIFactory(env=env, engine=engine)
engine.service = service
with service() as api:
workflow_id = create_workflow(api, source=TEMPLATE_DIR)
user_id = create_user(api)
with service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
run_id = start_run(api, group_id)
# Poll workflow state every second.
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
while run['state'] in st.ACTIVE_STATES:
time.sleep(1)
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
serialize.validate_run_handle(run, state=st.STATE_ERROR)
assert run['messages'][0] == 'some error'
def test_run_helloworld_sync_env(sync_service):
"""Successfully execute the helloworld example that contains a notebook step
in the Python environment that rund flowServ."""
# -- Setup ----------------------------------------------------------------
#
# Start a new run for the workflow template.
with sync_service() as api:
workflow_id = create_workflow(api,
source=BENCHMARK_DIR,
specfile=SPEC_FILE)
user_id = create_user(api)
with sync_service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
args = [serialize_arg('greeting', 'Hey there')]
run_id = start_run(api, group_id, arguments=args)
# -- Validate the run handle against the expected state -------------------
with sync_service(user_id=user_id) as api:
r = api.runs().get_run(run_id)
serialize.validate_run_handle(r, state=st.STATE_SUCCESS)
# The run should have the greetings.txt file as a result.
files = dict()
for obj in r['files']:
files[obj['name']] = obj['id']
assert len(files) == 2
fh = api.runs().get_result_file(run_id=run_id,
file_id=files['results/greetings.txt'])
value = fh.open().read().decode('utf-8').strip()
assert 'Hey there Alice!' in value
assert 'Hey there Bob!' in value
def test_cancel_remote_workflow(tmpdir):
"""Cancel the execution of a remote workflow."""
# -- Setup ----------------------------------------------------------------
#
env = Config().basedir(tmpdir)
engine = RemoteTestController(client=RemoteTestClient(runcount=100),
poll_interval=1,
is_async=True)
service = LocalAPIFactory(env=env, engine=engine)
engine.service = service
# -- Start a new run for the workflow template.
with service() as api:
workflow_id = create_workflow(api, source=TEMPLATE_DIR)
user_id = create_user(api)
with service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
run_id = start_run(api, group_id)
# -- Poll workflow state every second.
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
while run['state'] == st.STATE_PENDING:
time.sleep(1)
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
serialize.validate_run_handle(run, state=st.STATE_RUNNING)
with service(user_id=user_id) as api:
api.runs().cancel_run(run_id=run_id, reason='test')
# Sleep to ensure that the workflow monitor polls the state and makes an
# attempt to update the run state. This should raise an error for the
# monitor. The error is not propagated here or to the run.
time.sleep(3)
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
serialize.validate_run_handle(run, state=st.STATE_CANCELED)
assert run['messages'][0] == 'test'
def test_cancel_run_helloworld(async_service):
"""Test cancelling a helloworld run."""
# -- Setup ----------------------------------------------------------------
#
# Start a new run for the workflow template.
with async_service() as api:
workflow_id = create_workflow(api, source=BENCHMARK_DIR)
user_id = create_user(api)
with async_service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
names = io_file(data=['Alice', 'Bob', 'Zoe'], format='plain/text')
file_id = upload_file(api, group_id, names)
args = [
serialize_arg('names', serialize_fh(file_id)),
serialize_arg('sleeptime', 10),
serialize_arg('greeting', 'Hi')
]
run_id = start_run(api, group_id, arguments=args)
# Poll run after sleeping for one second.
time.sleep(1)
with async_service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
assert run['state'] in st.ACTIVE_STATES
# -- Cancel the active run ------------------------------------------------
with async_service(user_id=user_id) as api:
run = api.runs().cancel_run(
run_id=run_id,
reason='done'
)
assert run['state'] == st.STATE_CANCELED
assert run['messages'][0] == 'done'
with async_service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
assert run['state'] == st.STATE_CANCELED
assert run['messages'][0] == 'done'
def test_override_worker_config(sync_service, mock_docker):
"""Execute workflow with modified worker configuration."""
# Create workflow.
with sync_service() as api:
workflow_id = create_workflow(api,
source=TEMPLATE_DIR,
specfile=TEMPLATE_HELLOWORLD)
user_id = create_user(api)
# Start new run with default workers.
with sync_service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
run_id = start_run(api, group_id, arguments=dict())
with sync_service(user_id=user_id) as api:
r = api.runs().get_run(run_id)
assert r['state'] == state.STATE_ERROR
# Start new run with modified workers.
worker_config = {'workers': {'test': Docker()}}
with sync_service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
run_id = start_run(api,
group_id,
arguments=dict(),
config=worker_config)
with sync_service(user_id=user_id) as api:
r = api.runs().get_run(run_id)
assert r['state'] == state.STATE_SUCCESS
def test_run_remote_workflow_error(is_async, tmpdir):
"""Execute the remote workflow example synchronized and in asynchronous
mode when execution results in an error state.
"""
# -- Setup ----------------------------------------------------------------
env = Config().volume(FStore(basedir=str(tmpdir))).auth()
engine = RemoteWorkflowController(client=RemoteTestClient(
runcount=3, error='some error'),
poll_interval=0.1,
is_async=is_async)
service = LocalAPIFactory(env=env, engine=engine)
# Need to set the association between the engine and the service explicitly
# after the API is created.
engine.service = service
with service() as api:
workflow_id = create_workflow(api, source=BENCHMARK_DIR)
user_id = create_user(api)
with service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
# -- Unit test ------------------------------------------------------------
# Start a new run
with service(user_id=user_id) as api:
run_id = start_run(api, group_id)
# Poll workflow state every second.
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
watch_dog = 30
while run['state'] in st.ACTIVE_STATES and watch_dog:
time.sleep(1)
watch_dog -= 1
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
serialize.validate_run_handle(run, state=st.STATE_ERROR)
assert run['messages'] == ['some error']
def run_erroneous_workflow(service, specfile):
"""Execute the modified helloworld example."""
with service() as api:
# Create workflow template, user, and the workflow group.
workflow_id = create_workflow(
api,
source=TEMPLATE_DIR,
specfile=specfile
)
user_id = create_user(api)
with service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
# Upload the names file.
names = io_file(data=NAMES, format='txt/plain')
file_id = upload_file(api, group_id, names)
# Run the workflow.
arguments = [
serialize_arg('names', serialize_fh(file_id)),
serialize_arg('greeting', 'Hi')
]
run_id = start_run(api, group_id, arguments=arguments)
# Poll workflow state every second.
run = poll_run(service, run_id, user_id)
assert run['state'] == st.STATE_SUCCESS
with service() as api:
wh = api.workflows().get_workflow(workflow_id=workflow_id)
attmpts = 0
while 'postproc' not in wh:
time.sleep(1)
with service() as api:
wh = api.workflows().get_workflow(workflow_id=workflow_id)
attmpts += 1
if attmpts > 60:
break
assert 'postproc' in wh
serialize.validate_workflow_handle(wh)
attmpts = 0
while wh['postproc']['state'] in st.ACTIVE_STATES:
time.sleep(1)
with service() as api:
wh = api.workflows().get_workflow(workflow_id=workflow_id)
attmpts += 1
if attmpts > 60:
break
assert wh['postproc']['state'] not in st.ACTIVE_STATES
serialize.validate_workflow_handle(wh)
assert wh['postproc']['state'] == st.STATE_ERROR
def test_run_helloworld_async(async_service, target):
"""Execute the helloworld example."""
# -- Setup ----------------------------------------------------------------
#
# Start a new run for the workflow template.
with async_service() as api:
workflow_id = create_workflow(api, source=BENCHMARK_DIR)
user_id = create_user(api)
with async_service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
names = io_file(data=['Alice', 'Bob', 'Zoe'], format='plain/text')
file_id = upload_file(api, group_id, names)
args = [
serialize_arg('names', serialize_fh(file_id, target)),
serialize_arg('sleeptime', 1),
serialize_arg('greeting', 'Hi')
]
run_id = start_run(api, group_id, arguments=args)
# Poll workflow state every second.
with async_service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
watch_dog = 30
while run['state'] in st.ACTIVE_STATES and watch_dog:
time.sleep(1)
watch_dog -= 1
with async_service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
assert run['state'] == st.STATE_SUCCESS
files = dict()
for f in run['files']:
files[f['name']] = f['id']
fh = api.runs().get_result_file(
run_id=run_id,
file_id=files['results/greetings.txt']
)
greetings = fh.open().read().decode('utf-8').strip()
assert 'Hi Alice' in greetings
assert 'Hi Bob' in greetings
assert 'Hi Zoe' in greetings
fh = api.runs().get_result_file(
run_id=run_id,
file_id=files['results/analytics.json']
)
assert json.load(fh.open()) is not None
def test_cancel_remote_workflow(tmpdir):
"""Cancel the execution of a remote workflow."""
# -- Setup ----------------------------------------------------------------
env = Config().volume(FStore(basedir=str(tmpdir))).auth()
engine = RemoteWorkflowController(client=RemoteTestClient(runcount=100),
poll_interval=0.25,
is_async=True)
service = LocalAPIFactory(env=env, engine=engine)
# Need to set the association between the engine and the service explicitly
# after the API is created.
engine.service = service
with service() as api:
workflow_id = create_workflow(api, source=BENCHMARK_DIR)
user_id = create_user(api)
with service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
# -- Unit test ------------------------------------------------------------
# Start a new run
with service(user_id=user_id) as api:
run_id = start_run(api, group_id)
# -- Poll workflow state every second.
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
watch_dog = 30
while run['state'] == st.STATE_PENDING and watch_dog:
time.sleep(0.1)
watch_dog -= 1
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
serialize.validate_run_handle(run, state=st.STATE_RUNNING)
with service(user_id=user_id) as api:
api.runs().cancel_run(run_id=run_id, reason='test')
# Sleep to ensure that the workflow monitor polls the state and makes an
# attempt to update the run state. This should raise an error for the
# monitor. The error is not propagated here or to the run.
time.sleep(1)
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
serialize.validate_run_handle(run, state=st.STATE_CANCELED)
assert run['messages'][0] == 'test'
def test_run_remote_workflow(tmpdir, is_async):
"""Execute the remote workflow example synchronized and in asynchronous
mode.
"""
# -- Setup ----------------------------------------------------------------
#
env = Config().basedir(tmpdir)
engine = RemoteTestController(client=RemoteTestClient(runcount=3,
data=['success']),
poll_interval=1,
is_async=is_async)
service = LocalAPIFactory(env=env, engine=engine)
engine.service = service
# Start a new run for the workflow template.
with service() as api:
workflow_id = create_workflow(api, source=TEMPLATE_DIR)
user_id = create_user(api)
with service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
run_id = start_run(api, group_id)
# Poll workflow state every second.
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
watch_dog = 30
while run['state'] in st.ACTIVE_STATES and watch_dog:
time.sleep(1)
watch_dog -= 1
with service(user_id=user_id) as api:
run = api.runs().get_run(run_id=run_id)
serialize.validate_run_handle(run, state=st.STATE_SUCCESS)
files = dict()
for obj in run['files']:
files[obj['name']] = obj['id']
f_id = files['results/data.txt']
with service(user_id=user_id) as api:
fh = api.runs().get_result_file(run_id=run_id, file_id=f_id)
data = fh.open().read().decode('utf-8')
assert 'success' in data
def test_postproc_workflow(fsconfig, tmpdir):
"""Execute the modified helloworld example."""
# -- Setup ----------------------------------------------------------------
#
# It is important here that we do not use the SQLite in-memory database
# since this fails (for unknown reason; presumably due to different threads)
# when the post-processing run is updated.
# --
env = Config().basedir(tmpdir).run_async().auth()
env.update(fsconfig)
service = LocalAPIFactory(env=env)
# Start a new run for the workflow template.
with service() as api:
# Need to set the file store in the backend to the new instance as
# well. Otherwise, the post processing workflow may attempt to use
# the backend which was initialized prior with a different file store.
workflow_id = create_workflow(
api,
source=TEMPLATE_DIR,
specfile=SPEC_FILE
)
user_id = create_user(api)
# Create four groups and run the workflow with a slightly different input
# file
for i in range(4):
with service(user_id=user_id) as api:
group_id = create_group(api, workflow_id)
names = io_file(data=NAMES[:(i + 1)], format='plain/text')
file_id = upload_file(api, group_id, names)
# Set the template argument values
arguments = [
serialize_arg('names', serialize_fh(file_id)),
serialize_arg('greeting', 'Hi')
]
run_id = start_run(api, group_id, arguments=arguments)
# Poll workflow state every second.
run = poll_run(service, run_id, user_id)
assert run['state'] == st.STATE_SUCCESS
with service() as api:
wh = api.workflows().get_workflow(workflow_id=workflow_id)
attmpts = 0
while 'postproc' not in wh:
time.sleep(1)
with service() as api:
wh = api.workflows().get_workflow(workflow_id=workflow_id)
attmpts += 1
if attmpts > 60:
break
assert 'postproc' in wh
serialize.validate_workflow_handle(wh)
attmpts = 0
while wh['postproc']['state'] in st.ACTIVE_STATES:
time.sleep(1)
with service() as api:
wh = api.workflows().get_workflow(workflow_id=workflow_id)
attmpts += 1
if attmpts > 60:
break
serialize.validate_workflow_handle(wh)
with service() as api:
ranking = api.workflows().get_ranking(workflow_id=workflow_id)
serialize.validate_ranking(ranking)
for fobj in wh['postproc']['files']:
if fobj['name'] == 'results/compare.json':
file_id = fobj['id']
with service(user_id=user_id) as api:
fh = api.workflows().get_result_file(
workflow_id=workflow_id,
file_id=file_id
)
compare = util.read_object(fh.open())
assert len(compare) == (i + 1)
# Access the post-processing result files.
with service() as api:
fh = api.workflows().get_result_archive(workflow_id=workflow_id)
assert fh.name.startswith('run')
assert fh.mime_type == 'application/gzip'
