def testAssignerActor(self):
mock_scheduler_addr = '127.0.0.1:%d' % get_next_port()
with create_actor_pool(n_process=1,
backend='gevent',
address=mock_scheduler_addr) as pool:
pool.create_actor(ClusterInfoActor, [pool.cluster_info.address],
uid=ClusterInfoActor.default_name())
resource_ref = pool.create_actor(ResourceActor,
uid=ResourceActor.default_name())
chunk_meta_ref = pool.create_actor(
ChunkMetaActor, uid=ChunkMetaActor.default_name())
endpoint1 = 'localhost:12345'
endpoint2 = 'localhost:23456'
res = dict(hardware=dict(cpu=4, memory=4096))
def write_mock_meta():
resource_ref.set_worker_meta(endpoint1, res)
resource_ref.set_worker_meta(endpoint2, res)
g = gevent.spawn(write_mock_meta)
g.join()
assigner_ref = pool.create_actor(AssignerActor,
uid=AssignerActor.default_name())
session_id = str(uuid.uuid4())
chunk_key1 = str(uuid.uuid4())
chunk_key2 = str(uuid.uuid4())
chunk_key3 = str(uuid.uuid4())
op_info = {
'op_name': 'test_op',
'io_meta':
dict(input_chunks=[chunk_key1, chunk_key2, chunk_key3]),
'output_size': 512,
'retries': 0,
'optimize': {
'depth': 0,
'demand_depths': (),
'successor_size': 1,
'descendant_size': 0
}
}
chunk_meta_ref.set_chunk_meta(session_id,
chunk_key1,
size=512,
workers=(endpoint1, ))
chunk_meta_ref.set_chunk_meta(session_id,
chunk_key2,
size=512,
workers=(endpoint1, ))
chunk_meta_ref.set_chunk_meta(session_id,
chunk_key3,
size=512,
workers=(endpoint2, ))
workers = assigner_ref.get_worker_assignments(session_id, op_info)
self.assertEqual(workers[0], endpoint1)
def _prepare_test_graph(self, session_id, graph_key, mock_workers):
addr = '127.0.0.1:%d' % get_next_port()
a1 = mt.random.random((100,))
a2 = mt.random.random((100,))
s = a1 + a2
v1, v2 = mt.split(s, 2)
graph = DAG()
v1.build_graph(graph=graph, compose=False)
v2.build_graph(graph=graph, compose=False)
with create_actor_pool(n_process=1, backend='gevent', address=addr) as pool:
pool.create_actor(ClusterInfoActor, [pool.cluster_info.address],
uid=ClusterInfoActor.default_name())
resource_ref = pool.create_actor(ResourceActor, uid=ResourceActor.default_name())
pool.create_actor(ChunkMetaActor, uid=ChunkMetaActor.default_name())
pool.create_actor(AssignerActor, uid=AssignerActor.default_name())
graph_ref = pool.create_actor(GraphActor, session_id, graph_key, serialize_graph(graph),
uid=GraphActor.gen_name(session_id, graph_key))
for w in mock_workers:
resource_ref.set_worker_meta(w, dict(hardware=dict(cpu_total=4)))
graph_ref.prepare_graph()
graph_ref.analyze_graph()
graph_ref.create_operand_actors(_start=False)
yield pool, graph_ref
def testErrorOnPrepare(self, *_):
session_id = str(uuid.uuid4())
addr = '127.0.0.1:%d' % get_next_port()
with create_actor_pool(n_process=1, backend='gevent', address=addr) as pool:
pool.create_actor(ClusterInfoActor, [pool.cluster_info.address],
uid=ClusterInfoActor.default_name())
resource_ref = pool.create_actor(ResourceActor, uid=ResourceActor.default_name())
pool.create_actor(ChunkMetaActor, uid=ChunkMetaActor.default_name())
pool.create_actor(AssignerActor, uid=AssignerActor.default_name())
resource_ref.set_worker_meta('localhost:12345', dict(hardware=dict(cpu_total=4)))
resource_ref.set_worker_meta('localhost:23456', dict(hardware=dict(cpu_total=4)))
# error occurred in create_operand_actors
graph_key = str(uuid.uuid4())
expr = mt.random.random((8, 2), chunk_size=2) + 1
graph = expr.build_graph(compose=False)
serialized_graph = serialize_graph(graph)
graph_ref = pool.create_actor(GraphActor, session_id, graph_key, serialized_graph,
uid=GraphActor.gen_name(session_id, graph_key))
def _mock_raises(*_, **__):
raise RuntimeError
with patch_method(GraphActor.create_operand_actors, new=_mock_raises):
with self.assertRaises(RuntimeError):
graph_ref.execute_graph()
self.assertEqual(graph_ref.get_state(), GraphState.FAILED)
graph_ref.destroy()
# interrupted during create_operand_actors
graph_key = str(uuid.uuid4())
graph_ref = pool.create_actor(GraphActor, session_id, graph_key, serialized_graph,
uid=GraphActor.gen_name(session_id, graph_key))
def _mock_cancels(*_, **__):
graph_meta_ref = pool.actor_ref(GraphMetaActor.gen_name(session_id, graph_key))
graph_meta_ref.set_state(GraphState.CANCELLING)
with patch_method(GraphActor.create_operand_actors, new=_mock_cancels):
graph_ref.execute_graph()
self.assertEqual(graph_ref.get_state(), GraphState.CANCELLED)
# interrupted during previous steps
graph_key = str(uuid.uuid4())
graph_ref = pool.create_actor(GraphActor, session_id, graph_key, serialized_graph,
uid=GraphActor.gen_name(session_id, graph_key))
def _mock_cancels(*_, **__):
graph_meta_ref = pool.actor_ref(GraphMetaActor.gen_name(session_id, graph_key))
graph_meta_ref.set_state(GraphState.CANCELLING)
return dict()
with patch_method(GraphAnalyzer.calc_operand_assignments, new=_mock_cancels):
graph_ref.execute_graph()
self.assertEqual(graph_ref.get_state(), GraphState.CANCELLED)
def _run_operand_case(session_id, graph_key, tensor, execution_creator):
graph = tensor.build_graph(compose=False)
with create_actor_pool(n_process=1, backend='gevent') as pool:
pool.create_actor(ClusterInfoActor, [pool.cluster_info.address],
uid=ClusterInfoActor.default_name())
resource_ref = pool.create_actor(ResourceActor,
uid=ResourceActor.default_name())
pool.create_actor(ChunkMetaActor,
uid=ChunkMetaActor.default_name())
pool.create_actor(AssignerActor, uid=AssignerActor.default_name())
graph_ref = pool.create_actor(GraphActor,
session_id,
graph_key,
serialize_graph(graph),
uid=GraphActor.gen_name(
session_id, graph_key))
addr_dict = dict()
def _build_mock_ref(uid=None, address=None):
if address in addr_dict:
return addr_dict[address]
else:
r = addr_dict[address] = execution_creator(pool)
return r
# handle mock objects
OperandActor._get_raw_execution_ref.side_effect = _build_mock_ref
mock_resource = dict(hardware=dict(cpu=4, cpu_total=4, memory=512))
resource_ref.set_worker_meta('localhost:12345', mock_resource)
resource_ref.set_worker_meta('localhost:23456', mock_resource)
graph_ref.prepare_graph()
fetched_graph = graph_ref.get_chunk_graph()
graph_ref.analyze_graph()
final_keys = set()
for c in fetched_graph:
if fetched_graph.count_successors(c) == 0:
final_keys.add(c.op.key)
graph_ref.create_operand_actors()
graph_meta_ref = pool.actor_ref(
GraphMetaActor.gen_name(session_id, graph_key))
start_time = time.time()
while True:
pool.sleep(0.1)
if time.time() - start_time > 30:
raise SystemError('Wait for execution finish timeout')
if graph_meta_ref.get_state() in (GraphState.SUCCEEDED,
GraphState.FAILED,
GraphState.CANCELLED):
break
def testFailoverMessage(self):
mock_session_id = str(uuid.uuid4())
mock_graph_key = str(uuid.uuid4())
mock_chunk_key = str(uuid.uuid4())
addr = '127.0.0.1:%d' % get_next_port()
mock_worker_addr = '127.0.0.1:54132'
options.scheduler.worker_blacklist_time = 0.5
with create_actor_pool(n_process=1, backend='gevent', address=addr) as pool:
pool.create_actor(ClusterInfoActor, [pool.cluster_info.address],
uid=ClusterInfoActor.default_name())
pool.create_actor(AssignerActor, uid=AssignerActor.default_name())
session_manager_ref = pool.create_actor(
SessionManagerActor, uid=SessionManagerActor.default_name())
resource_ref = pool.create_actor(ResourceActor, uid=ResourceActor.default_name())
chunk_meta_ref = pool.create_actor(
ChunkMetaActor, uid=ChunkMetaActor.default_name())
session_ref = pool.actor_ref(session_manager_ref.create_session(mock_session_id))
chunk_meta_ref.set_chunk_meta(mock_session_id, mock_chunk_key,
size=80, shape=(10,), workers=(mock_worker_addr,))
with mock.patch(GraphActor.__module__ + '.' + GraphActor.__name__, new=MockGraphActor):
session_ref.submit_tensor_graph(None, mock_graph_key)
graph_ref = pool.actor_ref(GraphActor.gen_name(mock_session_id, mock_graph_key))
expire_time = time.time() - options.scheduler.status_timeout - 1
resource_ref.set_worker_meta(mock_worker_addr, dict(update_time=expire_time))
resource_ref.detect_dead_workers(_tell=True)
pool.sleep(0.2)
_, removes, lost_chunks = graph_ref.get_worker_change_args()
self.assertListEqual(removes, [mock_worker_addr])
self.assertListEqual(lost_chunks, [mock_chunk_key])
self.assertNotIn(mock_worker_addr, resource_ref.get_workers_meta())
resource_ref.set_worker_meta(mock_worker_addr, dict(update_time=time.time()))
self.assertNotIn(mock_worker_addr, resource_ref.get_workers_meta())
pool.sleep(0.4)
resource_ref.set_worker_meta(mock_worker_addr, dict(update_time=time.time()))
self.assertIn(mock_worker_addr, resource_ref.get_workers_meta())
def execute_case():
pool.create_actor(ClusterInfoActor,
[pool.cluster_info.address],
uid=ClusterInfoActor.default_name())
resource_ref = pool.create_actor(
ResourceActor, uid=ResourceActor.default_name())
pool.create_actor(ChunkMetaActor,
uid=ChunkMetaActor.default_name())
pool.create_actor(AssignerActor,
uid=AssignerActor.default_name())
graph_ref = pool.create_actor(GraphActor,
session_id,
graph_key,
serialize_graph(graph),
uid=GraphActor.gen_name(
session_id, graph_key))
addr_dict = dict()
def _build_mock_ref(uid=None, address=None):
if address in addr_dict:
return addr_dict[address]
else:
r = addr_dict[address] = pool.create_actor(
FakeExecutionActor, sleep=1)
return r
# handle mock objects
OperandActor._get_raw_execution_ref.side_effect = _build_mock_ref
mock_resource = dict(
hardware=dict(cpu=4, cpu_total=4, memory=512))
def write_mock_meta():
resource_ref.set_worker_meta('localhost:12345',
mock_resource)
resource_ref.set_worker_meta('localhost:23456',
mock_resource)
v = gevent.spawn(write_mock_meta)
v.join()
graph_ref.prepare_graph()
fetched_graph = graph_ref.get_chunk_graph()
graph_ref.scan_node()
graph_ref.place_initial_chunks()
final_keys = set()
for c in fetched_graph:
if fetched_graph.count_successors(c) == 0:
final_keys.add(c.op.key)
graph_ref.create_operand_actors()
graph_meta_ref = pool.actor_ref(
GraphMetaActor.gen_name(session_id, graph_key))
start_time = time.time()
cancel_called = False
while True:
gevent.sleep(0.1)
if not cancel_called and time.time() > start_time + 0.8:
cancel_called = True
graph_ref.stop_graph(_tell=True)
if time.time() - start_time > 30:
raise SystemError('Wait for execution finish timeout')
if graph_meta_ref.get_state() in (GraphState.SUCCEEDED,
GraphState.FAILED,
GraphState.CANCELLED):
break
def testOperandActorWithCancel(self, *_):
import logging
logging.basicConfig(level=logging.DEBUG)
arr = mt.random.randint(10, size=(10, 8), chunk_size=4)
arr_add = mt.random.randint(10, size=(10, 8), chunk_size=4)
arr2 = arr + arr_add
session_id = str(uuid.uuid4())
graph_key = str(uuid.uuid4())
graph = arr2.build_graph(compose=False)
with create_actor_pool(n_process=1, backend='gevent') as pool:
pool.create_actor(ClusterInfoActor, [pool.cluster_info.address],
uid=ClusterInfoActor.default_name())
resource_ref = pool.create_actor(ResourceActor, uid=ResourceActor.default_name())
pool.create_actor(ChunkMetaActor, uid=ChunkMetaActor.default_name())
pool.create_actor(AssignerActor, uid=AssignerActor.default_name())
graph_ref = pool.create_actor(GraphActor, session_id, graph_key, serialize_graph(graph),
uid=GraphActor.gen_name(session_id, graph_key))
addr_dict = dict()
def _build_mock_ref(uid=None, address=None):
if address in addr_dict:
return addr_dict[address]
else:
r = addr_dict[address] = pool.create_actor(
FakeExecutionActor, exec_delay=0.2)
return r
# handle mock objects
OperandActor._get_raw_execution_ref.side_effect = _build_mock_ref
mock_resource = dict(hardware=dict(cpu=4, cpu_total=4, memory=512))
for idx in range(20):
resource_ref.set_worker_meta('localhost:%d' % (idx + 12345), mock_resource)
graph_ref.prepare_graph(compose=False)
fetched_graph = graph_ref.get_chunk_graph()
graph_ref.analyze_graph()
final_keys = set()
for c in fetched_graph:
if fetched_graph.count_successors(c) == 0:
final_keys.add(c.op.key)
graph_ref.create_operand_actors()
graph_meta_ref = pool.actor_ref(GraphMetaActor.gen_name(session_id, graph_key))
start_time = time.time()
cancel_called = False
while True:
pool.sleep(0.05)
if not cancel_called and time.time() > start_time + 0.3:
cancel_called = True
graph_ref.stop_graph(_tell=True)
if time.time() - start_time > 30:
raise SystemError('Wait for execution finish timeout')
if graph_meta_ref.get_state() in (GraphState.SUCCEEDED, GraphState.FAILED, GraphState.CANCELLED):
break
self.assertEqual(graph_meta_ref.get_state(), GraphState.CANCELLED)
def prepare_graph_in_pool(self, expr, clean_io_meta=True, compose=False):
session_id = str(uuid.uuid4())
graph_key = str(uuid.uuid4())
graph = expr.build_graph(compose=compose)
serialized_graph = serialize_graph(graph)
chunked_graph = expr.build_graph(compose=compose, tiled=True)
addr = '127.0.0.1:%d' % get_next_port()
with create_actor_pool(n_process=1, backend='gevent', address=addr) as pool:
pool.create_actor(ClusterInfoActor, [pool.cluster_info.address],
uid=ClusterInfoActor.default_name())
resource_ref = pool.create_actor(ResourceActor, uid=ResourceActor.default_name())
pool.create_actor(ChunkMetaActor, uid=ChunkMetaActor.default_name())
pool.create_actor(AssignerActor, uid=AssignerActor.default_name())
graph_ref = pool.create_actor(GraphActor, session_id, graph_key, serialized_graph,
uid=GraphActor.gen_name(session_id, graph_key))
graph_ref.prepare_graph(compose=compose)
fetched_graph = graph_ref.get_chunk_graph()
self.assertIsNotNone(fetched_graph)
self.assertEqual(len(chunked_graph), len(fetched_graph))
graph_ref.analyze_graph(do_placement=False)
op_infos = graph_ref.get_operand_info()
for n in fetched_graph:
depth = op_infos[n.op.key]['optimize']['depth']
self.assertIsNotNone(depth)
successor_size = op_infos[n.op.key]['optimize']['successor_size']
self.assertIsNotNone(successor_size)
descendant_size = op_infos[n.op.key]['optimize']['descendant_size']
self.assertIsNotNone(descendant_size)
resource_ref.set_worker_meta('localhost:12345', dict(hardware=dict(cpu_total=4)))
resource_ref.set_worker_meta('localhost:23456', dict(hardware=dict(cpu_total=4)))
graph_ref.analyze_graph()
op_infos = graph_ref.get_operand_info()
for n in fetched_graph:
if fetched_graph.count_predecessors(n) != 0:
continue
target_worker = op_infos[n.op.key]['target_worker']
self.assertIsNotNone(target_worker)
graph_ref.create_operand_actors(_clean_io_meta=clean_io_meta)
op_infos = graph_ref.get_operand_info()
if not clean_io_meta:
orig_metas = dict()
for n in fetched_graph:
try:
meta = orig_metas[n.op.key]
except KeyError:
meta = orig_metas[n.op.key] = dict(
predecessors=set(), successors=set(), input_chunks=set(), chunks=set()
)
meta['predecessors'].update([pn.op.key for pn in fetched_graph.iter_predecessors(n)])
meta['successors'].update([sn.op.key for sn in fetched_graph.iter_successors(n)])
meta['input_chunks'].update([pn.key for pn in fetched_graph.iter_predecessors(n)])
meta['chunks'].update([c.key for c in n.op.outputs])
for n in fetched_graph:
self.assertEqual(op_infos[n.op.key]['op_name'], type(n.op).__name__)
io_meta = op_infos[n.op.key]['io_meta']
orig_io_meta = orig_metas[n.op.key]
self.assertSetEqual(set(io_meta['predecessors']), set(orig_io_meta['predecessors']))
self.assertSetEqual(set(io_meta['successors']), set(orig_io_meta['successors']))
self.assertSetEqual(set(io_meta['input_chunks']), set(orig_io_meta['input_chunks']))
self.assertSetEqual(set(io_meta['chunks']), set(orig_io_meta['chunks']))
yield pool, graph_ref