from __future__ import annotations
import asyncio
import inspect
import logging
import random
import sys
import weakref
from abc import ABC, abstractmethod
from contextlib import suppress
from typing import Any, ClassVar
import dask
from dask.utils import parse_timedelta
from distributed.comm import registry
from distributed.comm.addressing import parse_address
from distributed.metrics import time
from distributed.protocol.compression import get_default_compression
from distributed.protocol.pickle import HIGHEST_PROTOCOL
from distributed.utils import wait_for
logger = logging.getLogger(__name__)
class CommClosedError(IOError):
pass
class FatalCommClosedError(CommClosedError):
pass
[docs]class Comm(ABC):
"""
A message-oriented communication object, representing an established
communication channel. There should be only one reader and one
writer at a time: to manage current communications, even with a
single peer, you must create distinct ``Comm`` objects.
Messages are arbitrary Python objects. Concrete implementations
of this class can implement different serialization mechanisms
depending on the underlying transport's characteristics.
"""
_instances: ClassVar[weakref.WeakSet[Comm]] = weakref.WeakSet()
name: str | None
local_info: dict
remote_info: dict
handshake_options: dict
deserialize: bool
def __init__(self, deserialize: bool = True):
self._instances.add(self)
self.allow_offload = True # for deserialization in utils.from_frames
self.name = None
self.local_info = {}
self.remote_info = {}
self.handshake_options = {}
self.deserialize = deserialize
# XXX add set_close_callback()?
[docs] @abstractmethod
async def read(self, deserializers=None):
"""
Read and return a message (a Python object).
This method returns a coroutine.
Parameters
----------
deserializers : dict[str, tuple[Callable, Callable, bool]] | None
An optional dict appropriate for distributed.protocol.deserialize.
See :ref:`serialization` for more.
"""
[docs] @abstractmethod
async def write(self, msg, serializers=None, on_error=None):
"""
Write a message (a Python object).
This method returns a coroutine.
Parameters
----------
msg
on_error : str | None
The behavior when serialization fails. See
``distributed.protocol.core.dumps`` for valid values.
"""
[docs] @abstractmethod
async def close(self):
"""
Close the communication cleanly. This will attempt to flush
outgoing buffers before actually closing the underlying transport.
This method returns a coroutine.
"""
[docs] @abstractmethod
def abort(self):
"""
Close the communication immediately and abruptly.
Useful in destructors or generators' ``finally`` blocks.
"""
[docs] @abstractmethod
def closed(self):
"""Return whether the stream is closed."""
@property
@abstractmethod
def local_address(self) -> str:
"""The local address. For logging and debugging purposes only."""
@property
@abstractmethod
def peer_address(self) -> str:
"""The peer's address. For logging and debugging purposes only."""
@property
def extra_info(self):
"""
Return backend-specific information about the communication,
as a dict. Typically, this is information which is initialized
when the communication is established and doesn't vary afterwards.
"""
return {}
@staticmethod
def handshake_info():
return {
"compression": get_default_compression(),
"python": tuple(sys.version_info)[:3],
"pickle-protocol": HIGHEST_PROTOCOL,
}
@staticmethod
def handshake_configuration(local, remote):
try:
out = {
"pickle-protocol": min(
local["pickle-protocol"], remote["pickle-protocol"]
)
}
except KeyError as e:
raise ValueError(
"Your Dask versions may not be in sync. "
"Please ensure that you have the same version of dask "
"and distributed on your client, scheduler, and worker machines"
) from e
if local["compression"] == remote["compression"]:
out["compression"] = local["compression"]
else:
out["compression"] = None
return out
def __repr__(self):
return "<{}{} {} local={} remote={}>".format(
self.__class__.__name__,
" (closed)" if self.closed() else "",
self.name or "",
self.local_address,
self.peer_address,
)
[docs]class Listener(ABC):
[docs] @abstractmethod
async def start(self):
"""
Start listening for incoming connections.
"""
[docs] @abstractmethod
def stop(self):
"""
Stop listening. This does not shutdown already established
communications, but prevents accepting new ones.
"""
@property
@abstractmethod
def listen_address(self):
"""
The listening address as a URI string.
"""
@property
@abstractmethod
def contact_address(self):
"""
An address this listener can be contacted on. This can be
different from `listen_address` if the latter is some wildcard
address such as 'tcp://0.0.0.0:123'.
"""
async def __aenter__(self):
await self.start()
return self
async def __aexit__(self, exc_type, exc_value, traceback):
future = self.stop()
if inspect.isawaitable(future):
await future
def __await__(self):
async def _():
await self.start()
return self
return _().__await__()
async def on_connection(
self, comm: Comm, handshake_overrides: dict[str, Any] | None = None
) -> None:
local_info = {**comm.handshake_info(), **(handshake_overrides or {})}
timeout = dask.config.get("distributed.comm.timeouts.connect")
timeout = parse_timedelta(timeout, default="seconds")
try:
# Timeout is to ensure that we'll terminate connections eventually.
# Connector side will employ smaller timeouts and we should only
# reach this if the comm is dead anyhow.
await wait_for(comm.write(local_info), timeout=timeout)
handshake = await wait_for(comm.read(), timeout=timeout)
# This would be better, but connections leak if worker is closed quickly
# write, handshake = await asyncio.gather(comm.write(local_info), comm.read())
except Exception as e:
with suppress(Exception):
await comm.close()
raise CommClosedError(f"Comm {comm!r} closed.") from e
comm.remote_info = handshake
comm.remote_info["address"] = comm.peer_address
comm.local_info = local_info
comm.local_info["address"] = comm.local_address
comm.handshake_options = comm.handshake_configuration(
comm.local_info, comm.remote_info
)
class Connector(ABC):
@abstractmethod
async def connect(self, address, deserialize=True):
"""
Connect to the given address and return a Comm object.
This function returns a coroutine. It may raise EnvironmentError
if the other endpoint is unreachable or unavailable. It
may raise ValueError if the address is malformed.
"""
[docs]async def connect(
addr, timeout=None, deserialize=True, handshake_overrides=None, **connection_args
):
"""
Connect to the given address (a URI such as ``tcp://127.0.0.1:1234``)
and yield a ``Comm`` object. If the connection attempt fails, it is
retried until the *timeout* is expired.
"""
if timeout is None:
timeout = dask.config.get("distributed.comm.timeouts.connect")
timeout = parse_timedelta(timeout, default="seconds")
scheme, loc = parse_address(addr)
backend = registry.get_backend(scheme)
connector = backend.get_connector()
comm = None
start = time()
def time_left():
deadline = start + timeout
return max(0, deadline - time())
backoff_base = 0.01
attempt = 0
logger.debug("Establishing connection to %s", loc)
# Prefer multiple small attempts than one long attempt. This should protect
# primarily from DNS race conditions
# gh3104, gh4176, gh4167
intermediate_cap = timeout / 5
active_exception = None
while time_left() > 0:
try:
comm = await wait_for(
connector.connect(loc, deserialize=deserialize, **connection_args),
timeout=min(intermediate_cap, time_left()),
)
break
except FatalCommClosedError:
raise
# Note: CommClosed inherits from OSError
except (asyncio.TimeoutError, OSError) as exc:
active_exception = exc
# As described above, the intermediate timeout is used to distributed
# initial, bulk connect attempts homogeneously. In particular with
# the jitter upon retries we should not be worred about overloading
# any more DNS servers
intermediate_cap = timeout
# FullJitter see https://aws.amazon.com/blogs/architecture/exponential-backoff-and-jitter/
upper_cap = min(time_left(), backoff_base * (2**attempt))
backoff = random.uniform(0, upper_cap)
attempt += 1
logger.debug(
"Could not connect to %s, waiting for %s before retrying", loc, backoff
)
await asyncio.sleep(backoff)
else:
raise OSError(
f"Timed out trying to connect to {addr} after {timeout} s"
) from active_exception
local_info = {
**comm.handshake_info(),
**(handshake_overrides or {}),
}
try:
# This would be better, but connections leak if worker is closed quickly
# write, handshake = await asyncio.gather(comm.write(local_info), comm.read())
handshake = await wait_for(comm.read(), time_left())
await wait_for(comm.write(local_info), time_left())
except Exception as exc:
with suppress(Exception):
await comm.close()
raise OSError(
f"Timed out during handshake while connecting to {addr} after {timeout} s"
) from exc
comm.remote_info = handshake
comm.remote_info["address"] = comm._peer_addr
comm.local_info = local_info
comm.local_info["address"] = comm._local_addr
comm.handshake_options = comm.handshake_configuration(
comm.local_info, comm.remote_info
)
return comm
[docs]def listen(addr, handle_comm, deserialize=True, **kwargs):
"""
Create a listener object with the given parameters. When its ``start()``
method is called, the listener will listen on the given address
(a URI such as ``tcp://0.0.0.0``) and call *handle_comm* with a
``Comm`` object for each incoming connection.
*handle_comm* can be a regular function or a coroutine.
"""
try:
scheme, loc = parse_address(addr, strict=True)
except ValueError:
if kwargs.get("ssl_context"):
addr = "tls://" + addr
else:
addr = "tcp://" + addr
scheme, loc = parse_address(addr, strict=True)
backend = registry.get_backend(scheme)
return backend.get_listener(loc, handle_comm, deserialize, **kwargs)