# tractor: structured concurrent "actors".
# Copyright 2018-eternity Tyler Goodlet.
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU Affero General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU Affero General Public License for more details.
# You should have received a copy of the GNU Affero General Public License
# along with this program. If not, see .
'''
The fundamental core machinery implementing every "actor"
including the process-local, or "python-interpreter (aka global)
singleton) `Actor` primitive(s) and its internal `trio` machinery
implementing the low level runtime system supporting the
discovery, communication, spawning, supervision and cancellation
of other actors in a hierarchincal process tree.
The runtime's main entry point: `async_main()` opens the top level
supervision and service `trio.Nursery`s which manage the tasks responsible
for running all lower level spawning, supervision and msging layers:
- lowlevel transport-protocol init and persistent connectivity on
top of `._ipc` primitives; the transport layer.
- bootstrapping of connection/runtime config from the spawning
parent (actor).
- starting and supervising IPC-channel msg processing loops around
tranport connections from parent/peer actors in order to deliver
SC-transitive RPC via scheduling of `trio` tasks.
- registration of newly spawned actors with the discovery sys.
Glossary:
--------
- tn: a `trio.Nursery` or "task nursery".
- an: an `ActorNursery` or "actor nursery".
- root: top/parent-most scope/task/process/actor (or other runtime
primitive) in a hierarchical tree.
- parent-ish: "higher-up" in the runtime-primitive hierarchy.
- child-ish: "lower-down" in the runtime-primitive hierarchy.
'''
from __future__ import annotations
from contextlib import (
ExitStack,
)
from functools import partial
import importlib
import importlib.util
import os
from pathlib import Path
from pprint import pformat
import signal
import sys
from typing import (
Any,
Callable,
Type,
TYPE_CHECKING,
)
import uuid
import textwrap
from types import ModuleType
import warnings
import trio
from trio._core import _run as trio_runtime
from trio import (
CancelScope,
Nursery,
TaskStatus,
)
from tractor.msg import (
MsgType,
NamespacePath,
Stop,
pretty_struct,
types as msgtypes,
)
from ..trionics import (
collapse_eg,
maybe_open_nursery,
)
from ..ipc import (
Channel,
# IPCServer, # causes cycles atm..
_server,
)
from ..discovery._addr import (
UnwrappedAddress,
Address,
# default_lo_addrs,
get_address_cls,
wrap_address,
)
from .._context import (
mk_context,
Context,
)
from ..log import get_logger
from .._exceptions import (
ContextCancelled,
InternalError,
ModuleNotExposed,
MsgTypeError,
unpack_error,
)
from ..devx import (
debug,
pformat as _pformat
)
from ..discovery._api import get_registry
from ._portal import Portal
from . import _state
from ..spawn import _mp_fixup_main
from ..spawn._mp_fixup_main import ParentMainData
from . import _rpc
if TYPE_CHECKING:
from ._supervise import ActorNursery # noqa
from trio._channel import MemoryChannelState
log = get_logger('tractor')
def _get_mod_abspath(module: ModuleType) -> Path:
return Path(module.__file__).absolute()
def get_mod_nsps2fps(mod_ns_paths: list[str]) -> dict[str, str]:
'''
Deliver a table of py module namespace-path-`str`s mapped to
their "physical" `.py` file paths in the file-sys.
'''
nsp2fp: dict[str, str] = {}
for nsp in mod_ns_paths:
mod: ModuleType = importlib.import_module(nsp)
nsp2fp[nsp] = str(_get_mod_abspath(mod))
return nsp2fp
_bp = False
class Actor:
'''
The fundamental "runtime" concurrency primitive.
An "actor" is the combination of a regular Python process
executing a `trio.run()` task tree, communicating with other
"actors" through "memory boundary portals": `Portal`, which
provide a high-level async API around IPC "channels" (`Channel`)
which themselves encapsulate various (swappable) network
transport protocols for sending msgs between said memory domains
(processes, hosts, non-GIL threads).
Each "actor" is `trio.run()` scheduled "runtime" composed of many
concurrent tasks in a single thread. The "runtime" tasks conduct
a slew of low(er) level functions to make it possible for message
passing between actors as well as the ability to create new
actors (aka new "runtimes" in new processes which are supervised
via an "actor-nursery" construct). Each task which sends messages
to a task in a "peer" actor (not necessarily a parent-child,
depth hierarchy) is able to do so via an "address", which maps
IPC connections across memory boundaries, and a task request id
which allows for per-actor tasks to send and receive messages to
specific peer-actor tasks with which there is an ongoing RPC/IPC
dialog.
'''
is_registrar: bool = False
@property
def is_arbiter(self) -> bool:
'''
Deprecated, use `.is_registrar`.
'''
warnings.warn(
'`Actor.is_arbiter` is deprecated.\n'
'Use `.is_registrar` instead.',
DeprecationWarning,
stacklevel=2,
)
return self.is_registrar
@property
def is_root(self) -> bool:
'''
This actor is the parent most in the tree?
'''
return _state.is_root_process()
msg_buffer_size: int = 2**6
# nursery placeholders filled in by `async_main()`,
# - after fork for subactors.
# - during boot for the root actor.
_root_tn: Nursery|None = None
_service_tn: Nursery|None = None
_ipc_server: _server.IPCServer|None = None
@property
def ipc_server(self) -> _server.IPCServer:
'''
The IPC transport-server for this actor; normally
a process-singleton.
'''
return self._ipc_server
# Information about `__main__` from parent
_parent_main_data: ParentMainData
_parent_chan_cs: CancelScope|None = None
_spawn_spec: msgtypes.SpawnSpec|None = None
# if started on ``asycio`` running ``trio`` in guest mode
_infected_aio: bool = False
# TODO: nursery tracking like `trio` does?
# _ans: dict[
# tuple[str, str],
# list[ActorNursery],
# ] = {}
# Process-global stack closed at end on actor runtime teardown.
# NOTE: this is currently an undocumented public api.
lifetime_stack: ExitStack = ExitStack()
def __init__(
self,
name: str,
uuid: str,
*,
enable_modules: list[str] | None = None,
loglevel: str|None = None,
registry_addrs: list[Address]|None = None,
spawn_method: str|None = None,
inherit_parent_main: bool = True,
arbiter_addr: UnwrappedAddress|None = None,
) -> None:
'''
This constructor is called in the parent actor **before** the spawning
phase (aka before a new process is executed).
'''
self._aid = msgtypes.Aid(
name=name,
uuid=uuid,
pid=os.getpid(),
)
self._task: trio.Task|None = None
# state
self._cancel_complete = trio.Event()
self._cancel_called_by: tuple[str, tuple]|None = None
self._cancel_called: bool = False
# retrieve and store parent `__main__` data which
# will be passed to children
self._parent_main_data: ParentMainData = _mp_fixup_main._mp_figure_out_main(
inherit_parent_main=inherit_parent_main,
)
# TODO? only add this when `is_debug_mode() == True` no?
# always include debugging tools module
enable_modules = list(enable_modules or [])
if _state.is_root_process():
enable_modules.append('tractor.devx.debug._tty_lock')
self.enable_modules: dict[str, str] = get_mod_nsps2fps(
mod_ns_paths=enable_modules,
)
self._mods: dict[str, ModuleType] = {}
self.loglevel: str = loglevel
if arbiter_addr is not None:
warnings.warn(
'`Actor(arbiter_addr=)` is now deprecated.\n'
'Use `registry_addrs: list[Address]` instead.',
DeprecationWarning,
stacklevel=2,
)
registry_addrs: list[Address] = [
wrap_address(arbiter_addr)
]
# marked by the process spawning backend at startup
# will be None for the parent most process started
# manually by the user (the "registrar")
self._spawn_method: str = spawn_method
# RPC state
self._ongoing_rpc_tasks = trio.Event()
self._ongoing_rpc_tasks.set()
self._rpc_tasks: dict[
tuple[Channel, str], # (chan, cid)
tuple[Context, Callable, trio.Event] # (ctx=>, fn(), done?)
] = {}
# map {actor uids -> Context}
self._contexts: dict[
tuple[
tuple[str, str], # .uid
str, # .cid
str, # .side
],
Context
] = {}
self._parent_chan: Channel|None = None
self._forkserver_info: tuple|None = None
# track each child/sub-actor in it's locally
# supervising nursery
self._actoruid2nursery: dict[
tuple[str, str], # sub-`Actor.uid`
ActorNursery|None,
] = {}
# when provided, init the registry addresses property from
# input via the validator.
self._reg_addrs: list[UnwrappedAddress] = []
if registry_addrs:
_state._runtime_vars['_registry_addrs'] = self.reg_addrs = [
addr.unwrap()
for addr in registry_addrs
]
@property
def aid(self) -> msgtypes.Aid:
'''
This process-singleton-actor's "unique actor ID" in struct form.
See the `tractor.msg.Aid` struct for details.
'''
return self._aid
@property
def name(self) -> str:
return self._aid.name
@property
def uid(self) -> tuple[str, str]:
'''
This process-singleton's "unique (cross-host) ID".
Delivered from the `.Aid.name/.uuid` fields as a `tuple` pair
and should be multi-host unique despite a large distributed
process plane.
'''
msg: str = (
f'`{type(self).__name__}.uid` is now deprecated.\n'
'Use the new `.aid: tractor.msg.Aid` (struct) instead '
'which also provides additional named (optional) fields '
'beyond just the `.name` and `.uuid`.'
)
warnings.warn(
msg,
DeprecationWarning,
stacklevel=2,
)
return (
self._aid.name,
self._aid.uuid,
)
@property
def pid(self) -> int:
return self._aid.pid
@property
def repr_state(self) -> str:
if self.cancel_complete:
return 'cancelled'
elif canceller := self.cancel_caller:
return f' and cancel-called by {canceller}'
else:
return 'running'
def pformat(
self,
ds: str = ': ',
indent: int = 0,
privates: bool = False,
) -> str:
fmtstr: str = f'|_id: {self.aid.reprol()!r}\n'
if privates:
aid_nest_prefix: str = '|_aid='
aid_field_repr: str = _pformat.nest_from_op(
input_op='',
text=pretty_struct.pformat(
struct=self.aid,
field_indent=2,
),
op_suffix='',
nest_prefix=aid_nest_prefix,
nest_indent=0,
)
fmtstr: str = f'{aid_field_repr}'
if rent_chan := self._parent_chan:
fmtstr += (
f"|_parent{ds}{rent_chan.aid.reprol()}\n"
)
server: _server.IPCServer = self.ipc_server
if server:
if privates:
server_repr: str = self._ipc_server.pformat(
privates=privates,
)
# create field ln as a key-header indented under
# and up to the section's key prefix.
# ^XXX if we were to indent `repr(Server)` to
# ': '
# _here_^
server_repr: str = _pformat.nest_from_op(
input_op='', # nest as sub-obj
op_suffix='',
text=server_repr,
)
fmtstr += (
f"{server_repr}"
)
else:
fmtstr += (
f'|_ipc: {server.repr_state!r}\n'
)
fmtstr += (
f'|_rpc: {len(self._rpc_tasks)} active tasks\n'
)
# TODO, actually fix the .repr_state impl/output?
# append ipc-ctx state summary
# ctxs: dict = self._contexts
# if ctxs:
# ctx_states: dict[str, int] = {}
# for ctx in self._contexts.values():
# ctx_state: str = ctx.repr_state
# cnt = ctx_states.setdefault(ctx_state, 0)
# ctx_states[ctx_state] = cnt + 1
# fmtstr += (
# f" ctxs{ds}{ctx_states}\n"
# )
# runtime-state
task_name: str = ''
if task := self._task:
task_name: str = task.name
fmtstr += (
# TODO, this just like ctx?
f'|_state: {self.repr_state!r}\n'
f' task: {task_name}\n'
f' loglevel: {self.loglevel!r}\n'
f' subactors_spawned: {len(self._actoruid2nursery)}\n'
)
if not _state.is_root_process():
fmtstr += f' spawn_method: {self._spawn_method!r}\n'
if privates:
fmtstr += (
# f' actoruid2nursery{ds}{self._actoruid2nursery}\n'
f' cancel_complete{ds}{self._cancel_complete}\n'
f' cancel_called_by_remote{ds}{self._cancel_called_by}\n'
f' cancel_called{ds}{self._cancel_called}\n'
)
if fmtstr:
fmtstr: str = textwrap.indent(
text=fmtstr,
prefix=' '*(1 + indent),
)
_repr: str = (
f'<{type(self).__name__}(\n'
f'{fmtstr}'
f')>\n'
)
if indent:
_repr: str = textwrap.indent(
text=_repr,
prefix=' '*indent,
)
return _repr
__repr__ = pformat
@property
def reg_addrs(self) -> list[UnwrappedAddress]:
'''
List of (socket) addresses for all known (and contactable)
registry-service actors in "unwrapped" (i.e. IPC interchange
wire-compat) form.
If you are looking for the "wrapped" address form, use
`.registry_addrs` instead.
'''
return self._reg_addrs
@reg_addrs.setter
def reg_addrs(
self,
addrs: list[UnwrappedAddress],
) -> None:
if not addrs:
log.warning(
'Empty registry address list is invalid:\n'
f'{addrs}'
)
return
self._reg_addrs = addrs
@property
def registry_addrs(self) -> list[Address]:
return [wrap_address(uw_addr)
for uw_addr in self.reg_addrs]
def load_modules(
self,
) -> None:
'''
Load explicitly enabled python modules from local fs after
process spawn.
Since this actor may be spawned on a different machine from
the original nursery we need to try and load the local module
code manually (presuming it exists).
'''
try:
if (
self._spawn_method == 'trio'
and
(parent_data := self._parent_main_data)
):
if 'init_main_from_name' in parent_data:
_mp_fixup_main._fixup_main_from_name(
parent_data['init_main_from_name'])
elif 'init_main_from_path' in parent_data:
_mp_fixup_main._fixup_main_from_path(
parent_data['init_main_from_path'])
status: str = 'Attempting to import enabled modules:\n'
modpath: str
filepath: str
for modpath, filepath in self.enable_modules.items():
# XXX append the allowed module to the python path which
# should allow for relative (at least downward) imports.
sys.path.append(os.path.dirname(filepath))
status += (
f'|_{modpath!r} -> {filepath!r}\n'
)
mod: ModuleType = importlib.import_module(modpath)
self._mods[modpath] = mod
if modpath == '__main__':
self._mods['__mp_main__'] = mod
log.runtime(status)
except ModuleNotFoundError:
# it is expected the corresponding `ModuleNotExposed` error
# will be raised later
log.error(
f"Failed to import {modpath} in {self.name}"
)
raise
# ?TODO, factor this meth-iface into a new `.rpc` subsys primitive?
# - _get_rpc_func(),
# - _deliver_ctx_payload(),
# - get_context(),
# - start_remote_task(),
# - cancel_rpc_tasks(),
# - _cancel_task(),
#
def _get_rpc_func(self, ns, funcname):
'''
Try to lookup and return a target RPC func from the
post-fork enabled module set.
'''
try:
return getattr(self._mods[ns], funcname)
except KeyError as err:
mne = ModuleNotExposed(*err.args)
if ns == '__main__':
modpath = '__name__'
else:
modpath = f"'{ns}'"
msg = (
"\n\nMake sure you exposed the target module, `{ns}`, "
"using:\n"
"ActorNursery.start_actor(, enable_modules=[{mod}])"
).format(
ns=ns,
mod=modpath,
)
mne.msg += msg
raise mne
# TODO: rename to `._deliver_payload()` since this handles
# more then just `result` msgs now obvi XD
async def _deliver_ctx_payload(
self,
chan: Channel,
cid: str,
msg: MsgType|MsgTypeError,
) -> None|bool:
'''
Push an RPC msg-payload to the local consumer peer-task's
queue.
'''
aid: msgtypes.Aid = chan.aid
assert aid, f"`chan.aid` can't be {aid}"
try:
ctx: Context = self._contexts[(
aid.uid,
cid,
# TODO: how to determine this tho?
# side,
)]
except KeyError:
report: str = (
'Ignoring invalid IPC msg!?\n'
f'Ctx seems to not/no-longer exist??\n'
f'\n'
f'<=? {aid.reprol()!r}\n'
f' |_{pretty_struct.pformat(msg)}\n'
)
match msg:
case Stop():
log.runtime(report)
case _:
log.warning(report)
return
# if isinstance(msg, MsgTypeError):
# return await ctx._deliver_bad_msg()
return await ctx._deliver_msg(msg)
def get_context(
self,
chan: Channel,
cid: str,
nsf: NamespacePath,
# TODO: support lookup by `Context.side: str` ?
# -> would allow making a self-context which might have
# certain special use cases where RPC isolation is wanted
# between 2 tasks running in the same process?
# => prolly needs some deeper though on the real use cases
# and whether or not such things should be better
# implemented using a `TaskManager` style nursery..
#
# side: str|None = None,
msg_buffer_size: int|None = None,
allow_overruns: bool = False,
) -> Context:
'''
Look-up (existing) or create a new
inter-actor-SC-linked task "context" (a `Context`) which
encapsulates the local RPC task's execution enviroment
around `Channel` relayed msg handling including,
- a dedicated `trio` cancel scope (`Context._scope`),
- a pair of IPC-msg-relay "feeder" mem-channels
(`Context._recv/send_chan`),
- and a "context id" (cid) unique to the task-pair
msging session's lifetime.
'''
# ?TODO, use Aid here as well?
actor_uid = chan.aid.uid
assert actor_uid
try:
ctx = self._contexts[(
actor_uid,
cid,
# side,
)]
log.debug(
f'Retreived cached IPC ctx for\n'
f'peer: {chan.aid.uid}\n'
f'cid:{cid}\n'
)
ctx._allow_overruns: bool = allow_overruns
# adjust buffer size if specified
state: MemoryChannelState = ctx._send_chan._state # type: ignore
if (
msg_buffer_size
and
state.max_buffer_size != msg_buffer_size
):
state.max_buffer_size = msg_buffer_size
except KeyError:
log.debug(
f'Allocate new IPC ctx for\n'
f'peer: {chan.aid.uid}\n'
f'cid: {cid}\n'
)
ctx = mk_context(
chan,
cid,
nsf=nsf,
msg_buffer_size=msg_buffer_size or self.msg_buffer_size,
_allow_overruns=allow_overruns,
)
self._contexts[(
actor_uid,
cid,
# side,
)] = ctx
return ctx
async def start_remote_task(
self,
chan: Channel,
nsf: NamespacePath,
kwargs: dict,
# determines `Context.side: str`
portal: Portal|None = None,
# IPC channel config
msg_buffer_size: int|None = None,
allow_overruns: bool = False,
load_nsf: bool = False,
ack_timeout: float = float('inf'),
) -> Context:
'''
Send a `'cmd'` msg to a remote actor, which requests the
start and schedule of a remote task-as-function's
entrypoint.
Synchronously validates the endpoint type and returns
a (caller side) `Context` that can be used to accept
delivery of msg payloads from the local runtime's
processing loop: `._rpc.process_messages()`.
'''
cid: str = str(uuid.uuid4())
assert chan.aid.uid
ctx = self.get_context(
chan=chan,
cid=cid,
nsf=nsf,
# side='caller',
msg_buffer_size=msg_buffer_size,
allow_overruns=allow_overruns,
)
ctx._portal = portal
if (
'self' in nsf
or
not load_nsf
):
ns, _, func = nsf.partition(':')
else:
# TODO: pass nsf directly over wire!
# -[ ] but, how to do `self:`??
ns, func = nsf.to_tuple()
msg = msgtypes.Start(
ns=ns,
func=func,
kwargs=kwargs,
uid=self.aid.uid, # <- !TODO use .aid!
cid=cid,
)
log.runtime(
'Sending RPC `Start`\n\n'
f'=> peer: {chan.aid.uid}\n'
f' |_ {ns}.{func}({kwargs})\n\n'
f'{pretty_struct.pformat(msg)}'
)
await chan.send(msg)
# NOTE wait on first `StartAck` response msg and validate;
# this should be immediate and does not (yet) wait for the
# remote child task to sync via `Context.started()`.
with trio.fail_after(ack_timeout):
first_msg: msgtypes.StartAck = await ctx._rx_chan.receive()
try:
functype: str = first_msg.functype
except AttributeError:
raise unpack_error(first_msg, chan)
if functype not in (
'asyncfunc',
'asyncgen',
'context',
):
raise ValueError(
f'Invalid `StartAck.functype: str = {first_msg!r}` ??'
)
ctx._remote_func_type = functype
return ctx
async def _from_parent(
self,
parent_addr: UnwrappedAddress|None,
) -> tuple[
Channel,
list[UnwrappedAddress]|None,
list[str]|None, # preferred tpts
]:
'''
Bootstrap this local actor's runtime config from its parent by
connecting back via the IPC transport, handshaking and then
`Channel.recv()`-ing seeded data.
'''
try:
# Connect back to the parent actor and conduct initial
# handshake. From this point on if we error, we
# attempt to ship the exception back to the parent.
chan = await Channel.from_addr(
addr=wrap_address(parent_addr)
)
assert isinstance(chan, Channel)
# init handshake: swap actor-IDs.
await chan._do_handshake(aid=self.aid)
accept_addrs: list[UnwrappedAddress]|None = None
if self._spawn_method in (
'trio',
'subint',
# `main_thread_forkserver` (and the future
# variant-2 `subint_forkserver`) parent-side
# sends a `SpawnSpec` over IPC just like the
# other two — fork child-side runtime is
# trio-native.
'main_thread_forkserver',
'subint_forkserver',
):
# Receive post-spawn runtime state from our parent.
spawnspec: msgtypes.SpawnSpec = await chan.recv()
match spawnspec:
case MsgTypeError():
raise spawnspec
case msgtypes.SpawnSpec():
self._spawn_spec = spawnspec
log.runtime(
'Received runtime spec from parent:\n\n'
# TODO: eventually all these msgs as
# `msgspec.Struct` with a special mode that
# pformats them in multi-line mode, BUT only
# if "trace"/"util" mode is enabled?
f'{pretty_struct.pformat(spawnspec)}\n'
)
case _:
raise InternalError(
f'Received invalid non-`SpawnSpec` payload !?\n'
f'{spawnspec}\n'
)
# ^^XXX TODO XXX^^^
# when the `SpawnSpec` fails to decode the above will
# raise a `MsgTypeError` which if we do NOT ALSO
# RAISE it will tried to be pprinted in the
# log.runtime() below..
#
# SO we gotta look at how other `chan.recv()` calls
# are wrapped and do the same for this spec receive!
# -[ ] see `._rpc` likely has the answer?
# ^^^XXX NOTE XXX^^^, can't be called here!
#
# breakpoint()
# import pdbp; pdbp.set_trace()
#
# => bc we haven't yet received the
# `spawnspec._runtime_vars` which contains
# `debug_mode: bool`..
# `SpawnSpec.bind_addrs`
# ---------------------
accept_addrs: list[UnwrappedAddress] = spawnspec.bind_addrs
# `SpawnSpec._runtime_vars`
# -------------------------
# => update process-wide globals
# TODO! -[ ] another `Struct` for rtvs..
rvs: dict[str, Any] = spawnspec._runtime_vars
# `stackscope` SIGUSR1 handler: install when ANY of
# `_debug_mode` / `use_stackscope` rt-vars OR the
# `TRACTOR_ENABLE_STACKSCOPE` env var is set (the
# latter being a lighter test-time hang-debug path;
# see `tractor._testing.pytest`'s `--enable-stackscope`
# CLI flag — env var propagates via fork-inherited
# environ).
#
# NOTE, NOT *exclusively* gated on `_debug_mode` so
# SIGUSR1 task-tree dumps work in plain (non-pdb)
# runs too — but we DO still install under
# `_debug_mode` since otherwise the default SIGUSR1
# action would terminate the proc, esp. nasty in
# infected-`asyncio` sub-actors mid-REPL.
if (
rvs.get('_debug_mode')
or
rvs.get('use_stackscope')
or
os.environ.get('TRACTOR_ENABLE_STACKSCOPE')
):
from ..devx import enable_stack_on_sig
try:
# TODO: maybe return some status msgs upward
# to that we can emit them in `con_status`
# instead?
log.devx(
'Enabling `stackscope` traces on SIGUSR1'
)
enable_stack_on_sig()
except ImportError:
log.warning(
'`stackscope` not installed for use in '
'debug mode / `--enable-stackscope`!'
)
if rvs['_debug_mode']:
if rvs.get('use_greenback', False):
from ..devx import maybe_init_greenback
maybe_mod: ModuleType|None = await maybe_init_greenback()
if maybe_mod:
log.devx(
'Activated `greenback` '
'for `tractor.pause_from_sync()` support!'
)
else:
rvs['use_greenback'] = False
log.warning(
'`greenback` not installed for use in debug mode!\n'
'`tractor.pause_from_sync()` not available!'
)
# XXX ensure the "infected `asyncio` mode" setting
# passed down from our spawning parent is consistent
# with `trio`-runtime initialization:
# - during sub-proc boot, the entrypoint func
# (`._entry._main()`) should set
# `._infected_aio = True` before calling
# `run_as_asyncio_guest()`,
# - the value of `infect_asyncio: bool = True` as
# passed to `ActorNursery.start_actor()` must be
# the same as `_runtime_vars['_is_infected_aio']`
if (
(aio_rtv := rvs['_is_infected_aio'])
!=
(aio_attr := self._infected_aio)
):
raise InternalError(
'Parent sent runtime-vars that mismatch for the '
'"infected `asyncio` mode" settings ?!?\n\n'
f'rvs["_is_infected_aio"] = {aio_rtv}\n'
f'self._infected_aio = {aio_attr}\n'
)
if aio_rtv:
assert (
trio_runtime.GLOBAL_RUN_CONTEXT.runner.is_guest
# and
# ^TODO^ possibly add a `sniffio` or
# `trio` pub-API for `is_guest_mode()`?
)
rvs['_is_root'] = False # obvi XD
# TODO, remove! left in just while protoing init fix!
# global _bp
# if (
# 'chart' in self.aid.name
# and
# isinstance(
# rvs['_root_addrs'][0],
# dict,
# )
# and
# not _bp
# ):
# _bp = True
# breakpoint()
_state._runtime_vars.update(rvs)
# `SpawnSpec.reg_addrs`
# ---------------------
# => update parent provided registrar contact info
#
self.reg_addrs = [
# TODO: we don't really NEED these as tuples?
# so we can probably drop this casting since
# apparently in python lists are "more
# efficient"?
tuple(val)
for val in spawnspec.reg_addrs
]
# `SpawnSpec.enable_modules`
# ---------------------
# => extend RPC-python-module (capabilities) with
# those permitted by parent.
#
# NOTE, only the root actor should have
# a pre-permitted entry for `.devx.debug._tty_lock`.
assert not self.enable_modules
self.enable_modules.update(
spawnspec.enable_modules
)
self._parent_main_data = spawnspec._parent_main_data
# XXX QUESTION(s)^^^
# -[ ] already set in `.__init__()` right, but how is
# it diff from this blatant parent copy?
# -[ ] do we need/want the .__init__() value in
# just the root case orr?
return (
chan,
accept_addrs,
_state._runtime_vars['_enable_tpts']
)
# failed to connect back?
except (
OSError,
ConnectionError,
):
log.warning(
f'Failed to connect to spawning parent actor!?\n'
f'\n'
f'x=> {parent_addr}\n'
f' |_{self}\n\n'
)
await self.cancel(req_chan=None) # self cancel
raise
def cancel_soon(self) -> None:
'''
Cancel this actor asap; can be called from a sync context.
Schedules runtime cancellation via `Actor.cancel()` inside
the RPC service nursery.
'''
actor_repr: str = _pformat.nest_from_op(
input_op='>c(',
text=self.pformat(),
nest_indent=1,
)
log.cancel(
'Actor.cancel_soon()` was called!\n'
f'>> scheduling `Actor.cancel()`\n'
f'{actor_repr}'
)
assert self._service_tn
self._service_tn.start_soon(
self.cancel,
None, # self cancel all rpc tasks
)
# schedule a "canceller task" in the `._root_tn` once the
# `._service_tn` is fully shutdown; task waits for child-ish
# scopes to fully exit then finally cancels its parent,
# root-most, scope.
async def cancel_root_tn_after_services():
log.runtime(
'Waiting on service-tn to cancel..\n'
f'c>)\n'
f'|_{self._service_tn.cancel_scope!r}\n'
)
await self._cancel_complete.wait()
log.cancel(
f'`._service_tn` cancelled\n'
f'>c)\n'
f'|_{self._service_tn.cancel_scope!r}\n'
f'\n'
f'>> cancelling `._root_tn`\n'
f'c>(\n'
f' |_{self._root_tn.cancel_scope!r}\n'
)
self._root_tn.cancel_scope.cancel()
self._root_tn.start_soon(
cancel_root_tn_after_services
)
@property
def cancel_complete(self) -> bool:
return self._cancel_complete.is_set()
@property
def cancel_called(self) -> bool:
'''
Was this actor requested to cancel by a remote peer actor.
'''
return self._cancel_called_by is not None
@property
def cancel_caller(self) -> msgtypes.Aid|None:
return self._cancel_called_by
async def cancel(
self,
# chan whose lifetime limits the lifetime of its remotely
# requested and locally spawned RPC tasks - similar to the
# supervision semantics of a nursery wherein the actual
# implementation does start all such tasks in a sub-nursery.
req_chan: Channel|None,
) -> bool:
'''
Cancel this actor's runtime, eventually resulting in
termination of its containing OS process.
The ideal "deterministic" teardown sequence in order is:
- cancel all ongoing rpc tasks by cancel scope.
- cancel the channel server to prevent new inbound
connections.
- cancel the "service" nursery reponsible for
spawning new rpc tasks.
- return control the parent channel message loop.
'''
(
requesting_aid, # Aid
requester_type, # str
req_chan,
log_meth,
) = (
req_chan.aid,
'peer',
req_chan,
log.cancel,
) if req_chan else (
# a self cancel of ALL rpc tasks
self.aid,
'self',
self,
log.runtime,
)
# TODO: just use the new `Context.repr_rpc: str` (and
# other) repr fields instead of doing this all manual..
msg: str = (
f'Actor-runtime cancel request from {requester_type!r}\n'
f'\n'
f'<=c)\n'
f'{self}'
)
# TODO: what happens here when we self-cancel tho?
self._cancel_called_by: tuple = requesting_aid
self._cancel_called = True
# cancel all ongoing rpc tasks
with CancelScope(shield=True):
# kill any debugger request task to avoid deadlock
# with the root actor in this tree
debug_req = debug.DebugStatus
lock_req_ctx: Context = debug_req.req_ctx
if (
lock_req_ctx
and
lock_req_ctx.has_outcome
):
msg += (
f'\n'
f'-> Cancelling active debugger request..\n'
f'|_{debug.Lock.repr()}\n\n'
f'|_{lock_req_ctx}\n\n'
)
# lock_req_ctx._scope.cancel()
# TODO: wrap this in a method-API..
debug_req.req_cs.cancel()
# if lock_req_ctx:
# self-cancel **all** ongoing RPC tasks
await self.cancel_rpc_tasks(
req_aid=requesting_aid,
parent_chan=None,
)
# stop channel server
if ipc_server := self.ipc_server:
ipc_server.cancel()
await ipc_server.wait_for_shutdown()
# Break the shield on the parent-channel
# `process_messages` loop (started with `shield=True`
# in `async_main` above). Required to avoid a
# deadlock during teardown of fork-spawned subactors:
# without this cancel, the loop parks waiting for
# EOF on the parent channel, but the parent is
# blocked on `os.waitpid()` for THIS actor's exit
# — mutual wait. For exec-spawn backends the EOF
# arrives naturally when the parent closes its
# handler-task socket during its own teardown, but
# in fork backends the shared-process-image makes
# that delivery racy / not guaranteed. Explicit
# cancel here gives us deterministic unwinding
# regardless of backend.
if self._parent_chan_cs is not None:
self._parent_chan_cs.cancel()
# cancel all rpc tasks permanently
if self._service_tn:
self._service_tn.cancel_scope.cancel()
log_meth(msg)
self._cancel_complete.set()
return True
async def _cancel_task(
self,
cid: str,
parent_chan: Channel,
requesting_aid: msgtypes.Aid|None,
ipc_msg: dict|None|bool = False,
) -> bool:
'''
Cancel a local (RPC) task by context-id/channel by calling
`trio.CancelScope.cancel()` on it's surrounding cancel
scope.
'''
# this ctx based lookup ensures the requested task to be
# cancelled was indeed spawned by a request from its
# parent (or some grandparent's) channel
ctx: Context
func: Callable
is_complete: trio.Event
try:
(
ctx,
func,
is_complete,
) = self._rpc_tasks[(
parent_chan,
cid,
)]
scope: CancelScope = ctx._scope
except KeyError:
# NOTE: during msging race conditions this will often
# emit, some examples:
# - child returns a result before cancel-msg/ctxc-raised
# - child self raises ctxc before parent send request,
# - child errors prior to cancel req.
log.runtime(
'Cancel request for invalid RPC task.\n'
'The task likely already completed or was never started!\n\n'
f'<= canceller: {requesting_aid}\n'
f'=> {cid}@{parent_chan.aid.uid}\n'
f' |_{parent_chan}\n'
)
return True
log.cancel(
'Rxed cancel request for RPC task\n'
f'{ctx._task!r} <=c) {requesting_aid}\n'
f'|_>> {ctx.repr_rpc}\n'
# f'|_{ctx._task}\n'
# f' >> {ctx.repr_rpc}\n'
# f'=> {ctx._task}\n'
# f' >> Actor._cancel_task() => {ctx._task}\n'
# f' |_ {ctx._task}\n\n'
# TODO: better ascii repr for "supervisor" like
# a nursery or context scope?
# f'=> {parent_chan}\n'
# f' |_{ctx._task}\n'
# TODO: simplified `Context.__repr__()` fields output
# shows only application state-related stuff like,
# - ._stream
# - .closed
# - .started_called
# - .. etc.
# f' >> {ctx.repr_rpc}\n'
# f' |_ctx: {cid}\n'
# f' >> {ctx._nsf}()\n'
)
if (
ctx._canceller is None
and requesting_aid
):
ctx._canceller: tuple = requesting_aid.uid
# TODO: pack the RPC `{'cmd': }` msg into a ctxc and
# then raise and pack it here?
if (
ipc_msg
and ctx._cancel_msg is None
):
# assign RPC msg directly from the loop which usually
# the case with `ctx.cancel()` on the other side.
ctx._cancel_msg = ipc_msg
# don't allow cancelling this function mid-execution
# (is this necessary?)
if func is self._cancel_task:
log.error('Do not cancel a cancel!?')
return True
# TODO: shouldn't we eventually be calling ``Context.cancel()``
# directly here instead (since that method can handle both
# side's calls into it?
# await ctx.cancel()
scope.cancel()
# wait for _invoke to mark the task complete
flow_info: str = (
f'<= canceller: {requesting_aid}\n'
f'=> ipc-parent: {parent_chan}\n'
f'|_{ctx}\n'
)
log.runtime(
'Waiting on RPC task to cancel\n\n'
f'{flow_info}'
)
await is_complete.wait()
log.runtime(
f'Sucessfully cancelled RPC task\n\n'
f'{flow_info}'
)
return True
async def cancel_rpc_tasks(
self,
req_aid: msgtypes.Aid,
# NOTE: when None is passed we cancel **all** rpc
# tasks running in this actor!
parent_chan: Channel|None,
) -> None:
'''
Cancel all ongoing RPC tasks owned/spawned for a given
`parent_chan: Channel` or simply all tasks (inside
`._service_tn`) when `parent_chan=None`.
'''
tasks: dict = self._rpc_tasks
if not tasks:
log.runtime(
'Actor has no cancellable RPC tasks?\n'
f'<= canceller: {req_aid.reprol()}\n'
)
return
# TODO: seriously factor this into some helper funcs XD
tasks_str: str = ''
for (ctx, func, _) in tasks.values():
# TODO: std repr of all primitives in
# a hierarchical tree format, since we can!!
# like => repr for funcs/addrs/msg-typing:
#
# -[ ] use a proper utf8 "arm" like
# `stackscope` has!
# -[ ] for typed msging, show the
# py-type-annot style?
# - maybe auto-gen via `inspect` / `typing` type-sig:
# https://stackoverflow.com/a/57110117
# => see ex. code pasted into `.msg.types`
#
# -[ ] proper .maddr() for IPC primitives?
# - `Channel.maddr() -> str:` obvi!
# - `Context.maddr() -> str:`
tasks_str += (
f' |_@ /ip4/tcp/cid="{ctx.cid[-16:]} .."\n'
f' |>> {ctx._nsf}() -> dict:\n'
)
descr: str = (
'all' if not parent_chan
else
"IPC channel's "
)
rent_chan_repr: str = (
f' |_{parent_chan}\n\n'
if parent_chan
else ''
)
log.cancel(
f'Cancelling {descr} RPC tasks\n\n'
f'<=c) {req_aid} [canceller]\n'
f'{rent_chan_repr}'
f'c)=> {self.aid.uid} [cancellee]\n'
f' |_{self} [with {len(tasks)} tasks]\n'
# f' |_tasks: {len(tasks)}\n'
# f'{tasks_str}'
)
for (
(task_caller_chan, cid),
(ctx, func, is_complete),
) in tasks.copy().items():
if (
# maybe filter to specific IPC channel?
(parent_chan
and
task_caller_chan != parent_chan)
# never "cancel-a-cancel" XD
or (func == self._cancel_task)
):
continue
# TODO: this maybe block on the task cancellation
# and so should really done in a nursery batch?
await self._cancel_task(
cid,
task_caller_chan,
requesting_aid=req_aid,
)
if tasks:
log.cancel(
'Waiting for remaining rpc tasks to complete\n'
f'|_{tasks_str}'
)
await self._ongoing_rpc_tasks.wait()
@property
def accept_addrs(self) -> list[UnwrappedAddress]:
'''
All addresses to which the transport-channel server binds
and listens for new connections.
'''
return self._ipc_server.accept_addrs
@property
def accept_addr(self) -> UnwrappedAddress:
'''
Primary address to which the IPC transport server is
bound and listening for new connections.
'''
return self.accept_addrs[0]
# TODO, this should delegate ONLY to the
# `._spawn_spec._runtime_vars: dict` / `._state` APIs?
#
# XXX, AH RIGHT that's why..
# it's bc we pass this as a CLI flag to the child.py precisely
# bc we need the bootstrapping pre `async_main()`.. but maybe
# keep this as an impl deat and not part of the pub iface impl?
def is_infected_aio(self) -> bool:
'''
If `True`, this actor is running `trio` in guest mode on
the `asyncio` event loop and thus can use the APIs in
`.to_asyncio` to coordinate tasks running in each
framework but within the same actor runtime.
'''
return self._infected_aio
# ?TODO, is this the right type for this method?
def get_parent(self) -> Portal:
'''
Return a `Portal` to our parent.
'''
assert self._parent_chan, "No parent channel for this actor?"
return Portal(self._parent_chan)
# XXX: hard kill logic if needed?
# def _hard_mofo_kill(self):
# # If we're the root actor or zombied kill everything
# if self._parent_chan is None: # TODO: more robust check
# root = trio.lowlevel.current_root_task()
# for n in root.child_nurseries:
# n.cancel_scope.cancel()
async def async_main(
actor: Actor,
accept_addrs: UnwrappedAddress|None = None,
# XXX: currently ``parent_addr`` is only needed for the
# ``multiprocessing`` backend (which pickles state sent to
# the child instead of relaying it over the connect-back
# channel). Once that backend is removed we can likely just
# change this to a simple ``is_subactor: bool`` which will
# be False when running as root actor and True when as
# a subactor.
parent_addr: UnwrappedAddress|None = None,
task_status: TaskStatus[
tuple[
list[UnwrappedAddress], # accept_addrs
list[UnwrappedAddress], # reg_addrs
]
] = trio.TASK_STATUS_IGNORED,
) -> None:
'''
Main `Actor` runtime entrypoint; start the transport-specific
IPC channel server, (maybe) connect back to parent (to receive
additional config), startup all core `trio` machinery for
delivering RPCs, register with the discovery system.
The "root" (or "top-level") and "service" `trio.Nursery`s are
opened here and when cancelled/terminated effectively shutdown
the actor's "runtime" and all thus all ongoing RPC tasks.
'''
# XXX NOTE, `_state._current_actor` **must** be set prior to
# calling this core runtime entrypoint!
assert actor is _state.current_actor()
actor._task: trio.Task = trio.lowlevel.current_task()
# attempt to retreive ``trio``'s sigint handler and stash it
# on our debugger state.
debug.DebugStatus._trio_handler = signal.getsignal(signal.SIGINT)
is_registered: bool = False
try:
# establish primary connection with immediate parent
actor._parent_chan: Channel|None = None
# is this a sub-actor?
# get runtime info from parent.
if parent_addr is not None:
(
actor._parent_chan,
set_accept_addr_says_rent,
maybe_preferred_transports_says_rent,
) = await actor._from_parent(parent_addr)
accept_addrs: list[UnwrappedAddress] = []
# either it's passed in because we're not a child or
# because we're running in mp mode
if (
set_accept_addr_says_rent
and
set_accept_addr_says_rent is not None
):
accept_addrs = set_accept_addr_says_rent
else:
enable_transports: list[str] = (
maybe_preferred_transports_says_rent
or
[_state._def_tpt_proto]
)
for transport_key in enable_transports:
transport_cls: Type[Address] = get_address_cls(
transport_key
)
addr: Address = transport_cls.get_random()
accept_addrs.append(addr.unwrap())
# XXX, either passed in by caller or delivered
# in post spawn-spec handshake for subs.
if not accept_addrs:
raise RuntimeError(
f'No tpt bind addresses provided to actor!?\n'
f'parent_addr={parent_addr!r}\n'
f'accept_addrs={accept_addrs!r}\n'
f'enable_transports={enable_transports!r}\n'
)
ya_root_tn: bool = bool(actor._root_tn)
ya_service_tn: bool = bool(actor._service_tn)
# NOTE, a top-most "root" nursery in each actor-process
# enables a lifetime priority for the IPC-channel connection
# with a sub-actor's immediate parent. I.e. this connection
# is kept alive as a resilient service connection until all
# other machinery has exited, cancellation of all
# embedded/child scopes have completed. This helps ensure
# a deterministic (and thus "graceful")
# first-class-supervision style teardown where a parent actor
# (vs. say peers) is always the last to be contacted before
# disconnect.
root_tn: trio.Nursery
async with (
collapse_eg(),
maybe_open_nursery(
nursery=actor._root_tn,
) as root_tn,
):
if ya_root_tn:
assert root_tn is actor._root_tn
else:
actor._root_tn = root_tn
ipc_server: _server.IPCServer
async with (
collapse_eg(),
maybe_open_nursery(
nursery=actor._service_tn,
) as service_tn,
_server.open_ipc_server(
parent_tn=service_tn, # ?TODO, why can't this be the root-tn
stream_handler_tn=service_tn,
) as ipc_server,
):
if ya_service_tn:
assert service_tn is actor._service_tn
else:
# This nursery is used to handle all inbound
# connections to us such that if the TCP server
# is killed, connections can continue to process
# in the background until this nursery is cancelled.
actor._service_tn = service_tn
# set after allocate
actor._ipc_server = ipc_server
# load exposed/allowed RPC modules
# XXX: do this **after** establishing a channel to the parent
# but **before** starting the message loop for that channel
# such that import errors are properly propagated upwards
actor.load_modules()
# XXX TODO XXX: figuring out debugging of this
# would somemwhat guarantee "self-hosted" runtime
# debugging (since it hits all the ede cases?)
#
# `tractor.pause()` right?
# try:
# actor.load_modules()
# except ModuleNotFoundError as err:
# debug.pause_from_sync()
# import pdbp; pdbp.set_trace()
# raise
# Startup up the transport(-channel) server with,
# - subactor: the bind address is sent by our parent
# over our established channel
# - root actor: the ``accept_addr`` passed to this method
# TODO: why is this not with the root nursery?
# - see above that the `._service_tn` is what's used?
try:
eps: list = await ipc_server.listen_on(
accept_addrs=accept_addrs,
stream_handler_nursery=service_tn,
)
log.runtime(
f'Booted IPC server\n'
f'{ipc_server}\n'
)
assert (
(eps[0].listen_tn)
is not service_tn
)
except OSError as oserr:
# NOTE: always allow runtime hackers to debug
# tranport address bind errors - normally it's
# something silly like the wrong socket-address
# passed via a config or CLI Bo
entered_debug: bool = await debug._maybe_enter_pm(
oserr,
)
if not entered_debug:
log.exception('Failed to init IPC server !?\n')
else:
log.runtime('Exited debug REPL..')
raise
# TODO, just read direct from ipc_server?
accept_addrs: list[UnwrappedAddress] = actor.accept_addrs
# Register with the registrar if we're told its addr
log.runtime(
f'Registering `{actor.name}` => {pformat(accept_addrs)}\n'
# ^-TODO-^ we should instead show the maddr here^^
)
# TODO: ideally we don't fan out to all registrars
# if addresses point to the same actor..
# So we need a way to detect that? maybe iterate
# only on unique actor uids?
addr: UnwrappedAddress
for addr in actor.reg_addrs:
try:
waddr = wrap_address(addr)
assert waddr.is_valid
except AssertionError:
await debug.pause()
# !TODO, get rid of the local-portal crap XD
reg_portal: Portal
async with get_registry(addr) as reg_portal:
accept_addr: UnwrappedAddress
for accept_addr in accept_addrs:
accept_addr = wrap_address(accept_addr)
if not accept_addr.is_valid:
breakpoint()
await reg_portal.run_from_ns(
'self',
'register_actor',
uid=actor.aid.uid,
addr=accept_addr.unwrap(),
)
is_registered: bool = True
# init steps complete, deliver IPC-server and
# registrar addrs back to caller.
task_status.started((
accept_addrs,
actor.reg_addrs,
))
# Begin handling our new connection back to our
# parent. This is done last since we don't want to
# start processing parent requests until our channel
# server is 100% up and running.
if actor._parent_chan:
# Capture the shielded `loop_cs` for the
# parent-channel `process_messages` task so
# `Actor.cancel()` has a handle to break the
# shield during teardown — without this, the
# shielded loop would park on the parent chan
# indefinitely waiting for EOF that only arrives
# after the PARENT tears down, which under
# fork-based backends (e.g. `main_thread_forkserver`)
# it waits on THIS actor's exit — deadlock.
actor._parent_chan_cs = await root_tn.start(
partial(
_rpc.process_messages,
chan=actor._parent_chan,
shield=True,
)
)
log.runtime(
'Actor runtime is up!'
# 'Blocking on service nursery to exit..\n'
)
log.runtime(
'Service nursery complete\n'
'\n'
'->} waiting on root nursery to complete..\n'
)
# Blocks here as expected until the root nursery is
# killed (i.e. this actor is cancelled or signalled by the parent)
except Exception as internal_err:
if not is_registered:
err_report: str = (
'\n'
"Actor runtime (internally) failed BEFORE contacting the registry?\n"
f'registrars -> {actor.reg_addrs} ?!?!\n\n'
'^^^ THIS IS PROBABLY AN INTERNAL `tractor` BUG! ^^^\n\n'
'\t>> CALMLY CANCEL YOUR CHILDREN AND CALL YOUR PARENTS <<\n\n'
'\tIf this is a sub-actor hopefully its parent will keep running '
'and cancel/reap this sub-process..\n'
'(well, presuming this error was propagated upward)\n\n'
'\t---------------------------------------------\n'
'\tPLEASE REPORT THIS TRACEBACK IN A BUG REPORT @ ' # oneline
'https://github.com/goodboy/tractor/issues\n'
'\t---------------------------------------------\n'
)
# TODO: I guess we could try to connect back
# to the parent through a channel and engage a debugger
# once we have that all working with std streams locking?
log.exception(err_report)
if actor._parent_chan:
await _rpc.try_ship_error_to_remote(
actor._parent_chan,
internal_err,
)
# always!
match internal_err:
case ContextCancelled():
reprol: str = actor.aid.reprol()
log.cancel(
f'Actor {reprol!r} was task-ctx-cancelled with,\n'
f'\n'
f'{internal_err!r}'
)
case _:
log.exception(
'Main actor-runtime task errored\n'
f' Closing actor-lifetime-bound callbacks\n\n'
f'}}>\n'
f' |_{ls}\n'
f' |_{cbs_str}\n'
)
# XXX NOTE XXX this will cause an error which
# prevents any `infected_aio` actor from continuing
# and any callbacks in the `ls` here WILL NOT be
# called!!
# await debug.pause(shield=True)
ls.close()
# XXX TODO but hard XXX
# we can't actually do this bc the debugger uses the
# _service_tn to spawn the lock task, BUT, in theory if we had
# the root nursery surround this finally block it might be
# actually possible to debug THIS machinery in the same way
# as user task code?
#
# if actor.name == 'brokerd.ib':
# with CancelScope(shield=True):
# await debug.breakpoint()
# Unregister actor from the registry-sys / registrar.
if (
is_registered
and
not actor.is_registrar
):
failed_unreg: bool = False
rent_chan: Channel|None = actor._parent_chan
# XXX, detect whether the parent IS the registrar
# so we can FALL BACK to `_parent_chan` when a new
# connection attempt fails (e.g. UDS transport
# `os.unlink()`s the socket file during teardown).
#
# IMPORTANT: we do NOT eagerly reuse `_parent_chan`
# because it may still be carrying context/stream
# teardown protocol traffic — sending an
# `unregister_actor` RPC over it concurrently
# causes protocol-level conflicts. Instead we try
# a fresh `get_registry()` connection first and
# only fall back to the parent channel on failure.
#
# See `ipc._uds.close_listener()` for details on
# the UDS socket-file lifecycle.
parent_is_reg: bool = False
if (
rent_chan is not None
and
rent_chan.connected()
):
pchan_raddr: Address|None = rent_chan.raddr
if pchan_raddr is not None:
for reg_addr in actor.reg_addrs:
if (
pchan_raddr.unwrap()
==
tuple(reg_addr)
):
parent_is_reg = True
break
for addr in actor.reg_addrs:
waddr = wrap_address(addr)
assert waddr.is_valid
with trio.move_on_after(0.5) as cs:
cs.shield = True
try:
async with get_registry(
addr,
) as reg_portal:
await reg_portal.run_from_ns(
'self',
'unregister_actor',
uid=actor.aid.uid,
)
except OSError:
# Connection to registrar failed
# (listener socket likely already
# closed/unlinked). Fall back to
# parent channel if parent IS the
# registrar.
if (
parent_is_reg
and
rent_chan.connected()
):
try:
reg_portal = Portal(rent_chan)
await reg_portal.run_from_ns(
'self',
'unregister_actor',
uid=actor.aid.uid,
)
except (
OSError,
trio.ClosedResourceError,
):
failed_unreg = True
else:
failed_unreg = True
if cs.cancelled_caught:
failed_unreg = True
if failed_unreg:
teardown_report += (
f'-> Failed to unregister {actor.name} from '
f'registrar @ {addr}\n'
)
# Ensure all peers (actors connected to us as clients) are finished
if (
(ipc_server := actor.ipc_server)
and
ipc_server.has_peers(check_chans=True)
):
teardown_report += (
f'-> Waiting for remaining peers to clear..\n'
f' {pformat(ipc_server._peers)}'
)
log.runtime(teardown_report)
# NOTE: bound the peer-clear wait — otherwise if any
# peer-channel handler is stuck (e.g. never got its
# cancel propagated due to a runtime bug), this wait
# blocks forever and deadlocks the whole actor-tree
# teardown cascade. 3s is enough for any graceful
# cancel-ack round-trip; beyond that we're in bug
# territory and need to proceed with local teardown
# so the parent's `_ForkedProc.wait()` can unblock.
# See `ai/conc-anal/
# subint_forkserver_test_cancellation_leak_issue.md`
# for the full diagnosis.
with trio.move_on_after(3.0) as _peers_cs:
await ipc_server.wait_for_no_more_peers()
if _peers_cs.cancelled_caught:
teardown_report += (
f'-> TIMED OUT waiting for peers to clear '
f'({len(ipc_server._peers)} still connected)\n'
)
log.warning(teardown_report)
teardown_report += (
'-]> all peer channels are complete.\n'
)
# op_nested_actor_repr: str = _pformat.nest_from_op(
# input_op=')>',
# text=actor.pformat(),
# nest_prefix='|_',
# nest_indent=1, # under >
# )
teardown_report += (
'-)> actor runtime main task exit.\n'
# f'{op_nested_actor_repr}'
)
# if _state._runtime_vars['_is_root']:
# log.info(teardown_report)
# else:
log.runtime(teardown_report)
# Backward compat: class moved to discovery._registry
from ..discovery._registry import (
Registrar as Registrar,
)
Arbiter = Registrar