ray/python/ray/__init__.py
Clark Zinzow d958457d07
[Core] Second pass at privatizing APIs. (#17885)
* gcs_utils

* resource_spec

* profiling

* ray_perf and ray_cluster_perf

* test_utils
2021-08-18 20:56:33 -07:00

184 lines
5.8 KiB
Python

import os
import logging
logger = logging.getLogger(__name__)
def _configure_system():
import os
import platform
import sys
"""Wraps system configuration to avoid 'leaking' variables into ray."""
# MUST add pickle5 to the import path because it will be imported by some
# raylet modules.
if "pickle5" in sys.modules:
import pkg_resources
try:
version_info = pkg_resources.require("pickle5")
version = tuple(int(n) for n in version_info[0].version.split("."))
if version < (0, 0, 10):
raise ImportError("You are using an old version of pickle5 "
"that leaks memory, please run "
"'pip install pickle5 -U' to upgrade")
except pkg_resources.DistributionNotFound:
logger.warning("You are using the 'pickle5' module, but "
"the exact version is unknown (possibly carried as "
"an internal component by another module). Please "
"make sure you are using pickle5 >= 0.0.10 because "
"previous versions may leak memory.")
if "OMP_NUM_THREADS" not in os.environ:
logger.debug("[ray] Forcing OMP_NUM_THREADS=1 to avoid performance "
"degradation with many workers (issue #6998). You can "
"override this by explicitly setting OMP_NUM_THREADS.")
os.environ["OMP_NUM_THREADS"] = "1"
# Add the directory containing pickle5 to the Python path so that we find
# the pickle5 version packaged with ray and not a pre-existing pickle5.
pickle5_path = os.path.join(
os.path.abspath(os.path.dirname(__file__)), "pickle5_files")
sys.path.insert(0, pickle5_path)
# Importing psutil & setproctitle. Must be before ray._raylet is
# initialized.
thirdparty_files = os.path.join(
os.path.abspath(os.path.dirname(__file__)), "thirdparty_files")
sys.path.insert(0, thirdparty_files)
if sys.platform == "win32":
import ray._private.compat # noqa: E402
ray._private.compat.patch_redis_empty_recv()
if (platform.system() == "Linux"
and "Microsoft".lower() in platform.release().lower()):
import ray._private.compat # noqa: E402
ray._private.compat.patch_psutil()
# Expose ray ABI symbols which may be dependent by other shared
# libraries such as _streaming.so. See BUILD.bazel:_raylet
python_shared_lib_suffix = ".so" if sys.platform != "win32" else ".pyd"
so_path = os.path.join(
os.path.dirname(__file__), "_raylet" + python_shared_lib_suffix)
if os.path.exists(so_path):
import ctypes
from ctypes import CDLL
CDLL(so_path, ctypes.RTLD_GLOBAL)
_configure_system()
# Delete configuration function.
del _configure_system
# Replaced with the current commit when building the wheels.
__commit__ = "{{RAY_COMMIT_SHA}}"
__version__ = "2.0.0.dev0"
import ray._raylet # noqa: E402
from ray._raylet import ( # noqa: E402
ActorClassID, ActorID, NodeID, Config as _Config, JobID, WorkerID,
FunctionID, ObjectID, ObjectRef, TaskID, UniqueID, Language,
PlacementGroupID)
_config = _Config()
from ray.state import ( # noqa: E402
nodes, timeline, cluster_resources, available_resources,
)
from ray.worker import ( # noqa: E402,F401
LOCAL_MODE, SCRIPT_MODE, WORKER_MODE, RESTORE_WORKER_MODE,
UTIL_WORKER_MODE, SPILL_WORKER_MODE, cancel, get, get_actor, get_gpu_ids,
init, is_initialized, put, kill, remote, shutdown, wait,
)
import ray.internal # noqa: E402
# We import ray.actor because some code is run in actor.py which initializes
# some functions in the worker.
import ray.actor # noqa: E402,F401
from ray.actor import method # noqa: E402
from ray.cross_language import java_function, java_actor_class # noqa: E402
from ray.runtime_context import get_runtime_context # noqa: E402
from ray import data # noqa: E402,F401
from ray import util # noqa: E402
from ray import _private # noqa: E402,F401
# We import ClientBuilder so that modules can inherit from `ray.ClientBuilder`.
from ray.client_builder import client, ClientBuilder # noqa: E402
__all__ = [
"__version__",
"_config",
"get_runtime_context",
"actor",
"available_resources",
"cancel",
"client",
"ClientBuilder",
"cluster_resources",
"data"
"get",
"get_actor",
"get_gpu_ids",
"init",
"internal",
"is_initialized",
"java_actor_class",
"java_function",
"kill",
"Language",
"method",
"nodes",
"put",
"remote",
"shutdown",
"show_in_dashboard",
"timeline",
"util",
"wait",
"LOCAL_MODE",
"SCRIPT_MODE",
"WORKER_MODE",
]
# ID types
__all__ += [
"ActorClassID",
"ActorID",
"NodeID",
"JobID",
"WorkerID",
"FunctionID",
"ObjectID",
"ObjectRef",
"TaskID",
"UniqueID",
"PlacementGroupID",
]
# Add an alias so we can point to the final location in docs.
# TODO(yic) remove this once workflow is out of alpha.
from ray.experimental import workflow # noqa
__all__.append(workflow)
# Remove modules from top-level ray
def _ray_user_setup_function():
import os
user_setup_fn = os.environ.get("RAY_USER_SETUP_FUNCTION")
if user_setup_fn is not None:
try:
module_name, fn_name = user_setup_fn.rsplit(".", 1)
m = __import__(module_name, globals(), locals(), [fn_name])
getattr(m, fn_name)()
except Exception as e:
# We still need to allow ray to be imported, even there is
# something in the setup function.
logger.warning(
f"Failed to run user setup function: {user_setup_fn}. "
f"Error message {e}")
_ray_user_setup_function()
del os
del logging
del _ray_user_setup_function