ominfra 0.0.0.dev128__py3-none-any.whl → 0.0.0.dev129__py3-none-any.whl

ominfra/supervisor/supervisor.py

@@ -1,40 +1,38 @@
 # ruff: noqa: UP006 UP007
-import signal
+import errno
+import os
 import time
 import typing as ta
 
 from omlish.lite.check import check_isinstance
-from omlish.lite.check import check_not_none
 from omlish.lite.logs import log
 from omlish.lite.typing import Func1
 
 from .configs import ProcessGroupConfig
-from .context import ServerContextImpl
-from .dispatchers import Dispatchers
+from .configs import ServerConfig
 from .events import TICK_EVENTS
 from .events import EventCallbacks
 from .events import SupervisorRunningEvent
 from .events import SupervisorStoppingEvent
 from .groups import ProcessGroup
 from .groups import ProcessGroupManager
+from .io import IoManager
 from .poller import Poller
 from .process import PidHistory
 from .setup import SupervisorSetup
+from .signals import SignalHandler
 from .states import SupervisorState
-from .types import OutputDispatcher
+from .types import ExitNow
 from .types import Process
+from .types import SupervisorStateManager
 from .utils.os import decode_wait_status
-from .utils.signals import SignalReceiver
-from .utils.signals import sig_name
+from .utils.ostypes import Pid
+from .utils.ostypes import Rc
 
 
 ##
 
 
-class ExitNow(Exception):  # noqa
-    pass
-
-
 def timeslice(period: int, when: float) -> int:
     return int(when - (when % period))
 
@@ -42,59 +40,18 @@ def timeslice(period: int, when: float) -> int:
 ##
 
 
-class SignalHandler:
-    def __init__(
-            self,
-            *,
-            context: ServerContextImpl,
-            signal_receiver: SignalReceiver,
-            process_groups: ProcessGroupManager,
-    ) -> None:
+class SupervisorStateManagerImpl(SupervisorStateManager):
+    def __init__(self) -> None:
         super().__init__()
 
-        self._context = context
-        self._signal_receiver = signal_receiver
-        self._process_groups = process_groups
-
-    def set_signals(self) -> None:
-        self._signal_receiver.install(
-            signal.SIGTERM,
-            signal.SIGINT,
-            signal.SIGQUIT,
-            signal.SIGHUP,
-            signal.SIGCHLD,
-            signal.SIGUSR2,
-        )
-
-    def handle_signals(self) -> None:
-        sig = self._signal_receiver.get_signal()
-        if not sig:
-            return
-
-        if sig in (signal.SIGTERM, signal.SIGINT, signal.SIGQUIT):
-            log.warning('received %s indicating exit request', sig_name(sig))
-            self._context.set_state(SupervisorState.SHUTDOWN)
-
-        elif sig == signal.SIGHUP:
-            if self._context.state == SupervisorState.SHUTDOWN:
-                log.warning('ignored %s indicating restart request (shutdown in progress)', sig_name(sig))  # noqa
-            else:
-                log.warning('received %s indicating restart request', sig_name(sig))  # noqa
-                self._context.set_state(SupervisorState.RESTARTING)
+        self._state: SupervisorState = SupervisorState.RUNNING
 
-        elif sig == signal.SIGCHLD:
-            log.debug('received %s indicating a child quit', sig_name(sig))
-
-        elif sig == signal.SIGUSR2:
-            log.info('received %s indicating log reopen request', sig_name(sig))
-
-            for p in self._process_groups.all_processes():
-                for d in p.get_dispatchers():
-                    if isinstance(d, OutputDispatcher):
-                        d.reopen_logs()
+    @property
+    def state(self) -> SupervisorState:
+        return self._state
 
-        else:
-            log.debug('received %s indicating nothing', sig_name(sig))
+    def set_state(self, state: SupervisorState) -> None:
+        self._state = state
 
 
 ##
@@ -108,7 +65,7 @@ class Supervisor:
     def __init__(
             self,
             *,
-            context: ServerContextImpl,
+            config: ServerConfig,
             poller: Poller,
             process_groups: ProcessGroupManager,
             signal_handler: SignalHandler,
@@ -116,10 +73,12 @@ class Supervisor:
             process_group_factory: ProcessGroupFactory,
             pid_history: PidHistory,
             setup: SupervisorSetup,
+            states: SupervisorStateManager,
+            io: IoManager,
     ) -> None:
         super().__init__()
 
-        self._context = context
+        self._config = config
         self._poller = poller
         self._process_groups = process_groups
         self._signal_handler = signal_handler
@@ -127,6 +86,8 @@ class Supervisor:
         self._process_group_factory = process_group_factory
         self._pid_history = pid_history
         self._setup = setup
+        self._states = states
+        self._io = io
 
         self._ticks: ta.Dict[int, float] = {}
         self._stop_groups: ta.Optional[ta.List[ProcessGroup]] = None  # list used for priority ordered shutdown
@@ -136,11 +97,8 @@ class Supervisor:
     #
 
     @property
-    def context(self) -> ServerContextImpl:
-        return self._context
-
-    def get_state(self) -> SupervisorState:
-        return self._context.state
+    def state(self) -> SupervisorState:
+        return self._states.state
 
     #
 
@@ -181,7 +139,7 @@ class Supervisor:
                 log.info('waiting for %s to die', namestr)
                 self._last_shutdown_report = now
                 for proc in unstopped:
-                    log.debug('%s state: %s', proc.config.name, proc.get_state().name)
+                    log.debug('%s state: %s', proc.config.name, proc.state.name)
 
         return unstopped
 
@@ -205,7 +163,7 @@ class Supervisor:
         self._event_callbacks.clear()
 
         try:
-            for config in self._context.config.groups or []:
+            for config in self._config.groups or []:
                 self.add_process_group(config)
 
             self._signal_handler.set_signals()
@@ -229,7 +187,7 @@ class Supervisor:
         self._signal_handler.handle_signals()
         self._tick()
 
-        if self._context.state < SupervisorState.RUNNING:
+        if self._states.state < SupervisorState.RUNNING:
             self._ordered_stop_groups_phase_2()
 
     def _ordered_stop_groups_phase_1(self) -> None:
@@ -248,20 +206,11 @@ class Supervisor:
                 # down, so push it back on to the end of the stop group queue
                 self._stop_groups.append(group)
 
-    def get_dispatchers(self) -> Dispatchers:
-        return Dispatchers(
-            d
-            for p in self._process_groups.all_processes()
-            for d in p.get_dispatchers()
-        )
-
     def _poll(self) -> None:
-        dispatchers = self.get_dispatchers()
-
         sorted_groups = list(self._process_groups)
         sorted_groups.sort()
 
-        if self._context.state < SupervisorState.RUNNING:
+        if self._states.state < SupervisorState.RUNNING:
             if not self._stopping:
                 # first time, set the stopping flag, do a notification and set stop_groups
                 self._stopping = True
@@ -274,54 +223,7 @@ class Supervisor:
                 # if there are no unstopped processes (we're done killing everything), it's OK to shutdown or reload
                 raise ExitNow
 
-        for fd, dispatcher in dispatchers.items():
-            if dispatcher.readable():
-                self._poller.register_readable(fd)
-            if dispatcher.writable():
-                self._poller.register_writable(fd)
-
-        timeout = 1  # this cannot be fewer than the smallest TickEvent (5)
-        r, w = self._poller.poll(timeout)
-
-        for fd in r:
-            if fd in dispatchers:
-                try:
-                    dispatcher = dispatchers[fd]
-                    log.debug('read event caused by %r', dispatcher)
-                    dispatcher.handle_read_event()
-                    if not dispatcher.readable():
-                        self._poller.unregister_readable(fd)
-                except ExitNow:
-                    raise
-                except Exception:  # noqa
-                    dispatchers[fd].handle_error()
-            else:
-                # if the fd is not in combined map, we should unregister it. otherwise, it will be polled every
-                # time, which may cause 100% cpu usage
-                log.debug('unexpected read event from fd %r', fd)
-                try:
-                    self._poller.unregister_readable(fd)
-                except Exception:  # noqa
-                    pass
-
-        for fd in w:
-            if fd in dispatchers:
-                try:
-                    dispatcher = dispatchers[fd]
-                    log.debug('write event caused by %r', dispatcher)
-                    dispatcher.handle_write_event()
-                    if not dispatcher.writable():
-                        self._poller.unregister_writable(fd)
-                except ExitNow:
-                    raise
-                except Exception:  # noqa
-                    dispatchers[fd].handle_error()
-            else:
-                log.debug('unexpected write event from fd %r', fd)
-                try:
-                    self._poller.unregister_writable(fd)
-                except Exception:  # noqa
-                    pass
+        self._io.poll()
 
         for group in sorted_groups:
             for process in group:
@@ -331,17 +233,17 @@ class Supervisor:
         if depth >= 100:
             return
 
-        pid, sts = self._context.waitpid()
-        if not pid:
+        wp = waitpid()
+        if wp is None or not wp.pid:
             return
 
-        process = self._pid_history.get(pid, None)
+        process = self._pid_history.get(wp.pid, None)
         if process is None:
-            _, msg = decode_wait_status(check_not_none(sts))
-            log.info('reaped unknown pid %s (%s)', pid, msg)
+            _, msg = decode_wait_status(wp.sts)
+            log.info('reaped unknown pid %s (%s)', wp.pid, msg)
         else:
-            process.finish(check_not_none(sts))
-            del self._pid_history[pid]
+            process.finish(wp.sts)
+            del self._pid_history[wp.pid]
 
         if not once:
             # keep reaping until no more kids to reap, but don't recurse infinitely
@@ -366,3 +268,31 @@ class Supervisor:
             if this_tick != last_tick:
                 self._ticks[period] = this_tick
                 self._event_callbacks.notify(event(this_tick, self))
+
+
+##
+
+
+class WaitedPid(ta.NamedTuple):
+    pid: Pid
+    sts: Rc
+
+
+def waitpid() -> ta.Optional[WaitedPid]:
+    # Need pthread_sigmask here to avoid concurrent sigchld, but Python doesn't offer in Python < 3.4.  There is
+    # still a race condition here; we can get a sigchld while we're sitting in the waitpid call. However, AFAICT, if
+    # waitpid is interrupted by SIGCHLD, as long as we call waitpid again (which happens every so often during the
+    # normal course in the mainloop), we'll eventually reap the child that we tried to reap during the interrupted
+    # call. At least on Linux, this appears to be true, or at least stopping 50 processes at once never left zombies
+    # lying around.
+    try:
+        pid, sts = os.waitpid(-1, os.WNOHANG)
+    except OSError as exc:
+        code = exc.args[0]
+        if code not in (errno.ECHILD, errno.EINTR):
+            log.critical('waitpid error %r; a process may not be cleaned up properly', code)
+        if code == errno.EINTR:
+            log.debug('EINTR during reap')
+        return None
+    else:
+        return WaitedPid(pid, sts)  # type: ignore

ominfra/supervisor/types.py

@@ -5,7 +5,6 @@ import typing as ta
 
 from .configs import ProcessConfig
 from .configs import ProcessGroupConfig
-from .configs import ServerConfig
 from .states import ProcessState
 from .states import SupervisorState
 from .utils.collections import KeyedCollectionAccessors
@@ -21,6 +20,10 @@ if ta.TYPE_CHECKING:
 ##
 
 
+class ExitNow(Exception):  # noqa
+    pass
+
+
 ServerEpoch = ta.NewType('ServerEpoch', int)
 
 
@@ -44,12 +47,7 @@ class ConfigPriorityOrdered(abc.ABC):
 ##
 
 
-class ServerContext(abc.ABC):
-    @property
-    @abc.abstractmethod
-    def config(self) -> ServerConfig:
-        raise NotImplementedError
-
+class SupervisorStateManager(abc.ABC):
     @property
     @abc.abstractmethod
     def state(self) -> SupervisorState:
@@ -64,11 +62,6 @@ class ServerContext(abc.ABC):
 
 
 class Dispatcher(abc.ABC):
-    @property
-    @abc.abstractmethod
-    def process(self) -> 'Process':
-        raise NotImplementedError
-
     @property
     @abc.abstractmethod
     def channel(self) -> str:
@@ -112,8 +105,32 @@ class Dispatcher(abc.ABC):
     def handle_write_event(self) -> None:
         raise TypeError
 
+    #
+
+    def handle_connect(self) -> None:
+        raise TypeError
+
+    def handle_close(self) -> None:
+        raise TypeError
+
+    def handle_accepted(self, sock, addr) -> None:
+        raise TypeError
+
+
+class HasDispatchers(abc.ABC):
+    @abc.abstractmethod
+    def get_dispatchers(self) -> 'Dispatchers':
+        raise NotImplementedError
+
 
-class OutputDispatcher(Dispatcher, abc.ABC):
+class ProcessDispatcher(Dispatcher, abc.ABC):
+    @property
+    @abc.abstractmethod
+    def process(self) -> 'Process':
+        raise NotImplementedError
+
+
+class ProcessOutputDispatcher(ProcessDispatcher, abc.ABC):
     @abc.abstractmethod
     def remove_logs(self) -> None:
         raise NotImplementedError
@@ -123,7 +140,7 @@ class OutputDispatcher(Dispatcher, abc.ABC):
         raise NotImplementedError
 
 
-class InputDispatcher(Dispatcher, abc.ABC):
+class ProcessInputDispatcher(ProcessDispatcher, abc.ABC):
     @abc.abstractmethod
     def write(self, chars: ta.Union[bytes, str]) -> None:
         raise NotImplementedError
@@ -136,7 +153,11 @@ class InputDispatcher(Dispatcher, abc.ABC):
 ##
 
 
-class Process(ConfigPriorityOrdered, abc.ABC):
+class Process(
+    ConfigPriorityOrdered,
+    HasDispatchers,
+    abc.ABC,
+):
     @property
     @abc.abstractmethod
     def name(self) -> str:
@@ -159,11 +180,6 @@ class Process(ConfigPriorityOrdered, abc.ABC):
 
     #
 
-    @property
-    @abc.abstractmethod
-    def context(self) -> ServerContext:
-        raise NotImplementedError
-
     @abc.abstractmethod
     def finish(self, sts: Rc) -> None:
         raise NotImplementedError
@@ -180,18 +196,15 @@ class Process(ConfigPriorityOrdered, abc.ABC):
     def transition(self) -> None:
         raise NotImplementedError
 
+    @property
     @abc.abstractmethod
-    def get_state(self) -> ProcessState:
+    def state(self) -> ProcessState:
         raise NotImplementedError
 
     @abc.abstractmethod
     def after_setuid(self) -> None:
         raise NotImplementedError
 
-    @abc.abstractmethod
-    def get_dispatchers(self) -> 'Dispatchers':
-        raise NotImplementedError
-
 
 ##
 

{ominfra-0.0.0.dev128.dist-info → ominfra-0.0.0.dev129.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ominfra
-Version: 0.0.0.dev128
+Version: 0.0.0.dev129
 Summary: ominfra
 Author: wrmsr
 License: BSD-3-Clause
@@ -12,8 +12,8 @@ Classifier: Operating System :: OS Independent
 Classifier: Operating System :: POSIX
 Requires-Python: >=3.12
 License-File: LICENSE
-Requires-Dist: omdev==0.0.0.dev128
-Requires-Dist: omlish==0.0.0.dev128
+Requires-Dist: omdev==0.0.0.dev129
+Requires-Dist: omlish==0.0.0.dev129
 Provides-Extra: all
 Requires-Dist: paramiko~=3.5; extra == "all"
 Requires-Dist: asyncssh~=2.18; extra == "all"

{ominfra-0.0.0.dev128.dist-info → ominfra-0.0.0.dev129.dist-info}/RECORD

@@ -22,7 +22,7 @@ ominfra/clouds/aws/journald2aws/poster.py,sha256=hz1XuctW8GtLmfjhRvCFY6py52D4BzX
 ominfra/clouds/gcp/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ominfra/clouds/gcp/auth.py,sha256=3PyfRJNgajjMqJFem3SKui0CqGeHEsZlvbRhuxFcZG8,1348
 ominfra/deploy/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-ominfra/deploy/_executor.py,sha256=osVBYirgpX1iJKyLHf7JwKD3AhPqJrnCrjI4qVrNFJo,34662
+ominfra/deploy/_executor.py,sha256=1DQxeHJjxtRGxdFO7PiPhzxwiJlm8K2BelJfLXb_OxU,35079
 ominfra/deploy/configs.py,sha256=qi0kwT7G2NH7dXLOQic-u6R3yeadup_QtvrjwWIggbM,435
 ominfra/deploy/remote.py,sha256=6ACmpXU1uBdyGs3Xsp97ktKFq30cJlzN9LRWNUWlGY4,2144
 ominfra/deploy/executor/__init__.py,sha256=Y3l4WY4JRi2uLG6kgbGp93fuGfkxkKwZDvhsa0Rwgtk,15
@@ -56,37 +56,38 @@ ominfra/journald/tailer.py,sha256=5abcFMfgi7fnY9ZEQe2ZVobaJxjQkeu6d9Kagw33a1w,33
 ominfra/manage/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ominfra/manage/manage.py,sha256=BttL8LFEknHZE_h2Pt5dAqbfUkv6qy43WI0raXBZ1a8,151
 ominfra/pyremote/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-ominfra/pyremote/_runcommands.py,sha256=86bIzkhi5YLmtxQYyk8SBwPv8fCCfdu2NHHlxug4ACk,28498
+ominfra/pyremote/_runcommands.py,sha256=bzN-253qyoKwS-5OIqPmfhxVJdt4HtrnspeUbpKKPWM,28915
 ominfra/pyremote/bootstrap.py,sha256=RvMO3YGaN1E4sgUi1JEtiPak8cjvqtc_vRCq1yqbeZg,3370
 ominfra/pyremote/runcommands.py,sha256=bviS0_TDIoZVAe4h-_iavbvJtVSFu8lnk7fQ5iasCWE,1571
 ominfra/scripts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-ominfra/scripts/journald2aws.py,sha256=UpwV8P1ZE8ywXW-8xpCaTCBRWe94BuG_MS1CXfD3wcw,128547
-ominfra/scripts/supervisor.py,sha256=ZqUDlxRxSML5Ty0sOKuils35km5-9xjYBQF4yQDnVaY,222164
+ominfra/scripts/journald2aws.py,sha256=1fqTSLiZE3BCCxdyj5QMCWGHN4leS5INORaDuKyRm_A,128964
+ominfra/scripts/supervisor.py,sha256=y84cLJj98qvyHiwWf0bDly2Rwexvc9QaT0qrQx06lGc,224746
 ominfra/supervisor/LICENSE.txt,sha256=yvqaMNsDhWxziHa9ien6qCW1SkZv-DQlAg96XjfSee8,1746
 ominfra/supervisor/__init__.py,sha256=Y3l4WY4JRi2uLG6kgbGp93fuGfkxkKwZDvhsa0Rwgtk,15
 ominfra/supervisor/__main__.py,sha256=I0yFw-C08OOiZ3BF6lF1Oiv789EQXu-_j6whDhQUTEA,66
 ominfra/supervisor/configs.py,sha256=AhBlbifwDXc0acEhcbdv9jphJL-SBFODFDAWDVckzAE,3945
-ominfra/supervisor/context.py,sha256=UDJc0mVHHFg3T7MyO7gehWX1wwmFvw-7xeKLl-cC8jc,2595
-ominfra/supervisor/dispatchers.py,sha256=qWtVLx-4H3I-Io7OPcVgRpIYehGsA4q3Duo5ZWdUM4Y,970
-ominfra/supervisor/dispatchersimpl.py,sha256=t1VFcofj1kTH1q13Z-S1OUTXixPwgSqJkh5A5IkHKPA,10956
+ominfra/supervisor/dispatchers.py,sha256=8yReui-dVPfvqYrHLsCU7M0siluOs1p5Up7Ag4nMAeI,991
+ominfra/supervisor/dispatchersimpl.py,sha256=EZYxK_RbOkjMv4N3pjH6-_fBhXDmcxFhEyZEgEAbtUk,11294
 ominfra/supervisor/events.py,sha256=w3HQFrq-SuroYWoQfNFYeU1phnTvHTgsAqA6TGtAafI,6593
 ominfra/supervisor/exceptions.py,sha256=Qbu211H3CLlSmi9LsSikOwrcL5HgJP9ugvcKWlGTAoI,750
 ominfra/supervisor/groups.py,sha256=g5Zp_lkVhn1FSe6GSEPbaELincG5a46ctv1xpB-WmnQ,2163
-ominfra/supervisor/groupsimpl.py,sha256=_5Mxf6VdVV375KgwcuUTpErUH4FBpmzbNm3vbGB_JaQ,2304
-ominfra/supervisor/inject.py,sha256=ZCfloXxURjiVTutzudGFWybrVCvjDD_C3HayMATCMrY,3042
-ominfra/supervisor/main.py,sha256=RCCpMX4D2RKT8vZMF82pENKpXfW3GUdNgMEUzUtpqL0,4140
+ominfra/supervisor/groupsimpl.py,sha256=nIrW4SmB0W6c2jOR_HhkfVcH4eHyLZnG1FJ0MCzc6mQ,2292
+ominfra/supervisor/inject.py,sha256=J1XGUkgCuaD-MPASso-2sbTSoVuDDIhcGbazQLVjajY,3217
+ominfra/supervisor/io.py,sha256=iYYqQO0Hu1FIYzYXdSGHEIHYW-tGm4SCWky_ZTHaKWY,2727
+ominfra/supervisor/main.py,sha256=ebe7skFPfwXV2meMVRndhuLZmz-LiuHH1x1CgiarR0o,4132
 ominfra/supervisor/pipes.py,sha256=XrJ9lD04tPdzZD3xhhYKxpBKHWhZ0Ii315E78bgj7ws,2233
 ominfra/supervisor/poller.py,sha256=LnQVttPCm8a1UtnDvsho6zLw8NP-2_2VUiNM-d0w_FU,7776
 ominfra/supervisor/privileges.py,sha256=bO7rJGT7cMOBALK_4D4NiQnOS5dOYb14Sz66R-ymG24,2071
 ominfra/supervisor/process.py,sha256=UaubVxsxVqDnbuWVpTH0DTGbJGLO0vGJ9mNcvy2kCXM,217
-ominfra/supervisor/processimpl.py,sha256=0i0P3gdxuJsgcvy2VTKN6-NYgcBjozUaPsA_tUcEmb4,18769
+ominfra/supervisor/processimpl.py,sha256=bFhf9RqAB_yNlMM5dM1_4m964T9m3vKThEv3n7OYQN4,18684
 ominfra/supervisor/setup.py,sha256=7HwwwI-WT_Z0WjZ9_l5Orr4K298nKKhQ1f_ZgGsi9TU,622
 ominfra/supervisor/setupimpl.py,sha256=S_YgCH3XzLsFIAriJROvDMUDh7OzVVJoxzEzCkbb4g4,9648
+ominfra/supervisor/signals.py,sha256=jY52naUifcAjd6nICTP1ZW3IQSPsHB4cvbsJo8_QV_U,2196
 ominfra/supervisor/spawning.py,sha256=i1k3tmqWyU-KIN7kel-JVxTVGnLiTIVmZzlstJSZpjM,622
-ominfra/supervisor/spawningimpl.py,sha256=0InDyRPh7gAEX07lg6eUYMymX0RikHOz_ieAghxKx8Y,11063
+ominfra/supervisor/spawningimpl.py,sha256=E702F10TexP5-V6i97U6ysiBcTKyLSPzQoJZtedUjQs,11140
 ominfra/supervisor/states.py,sha256=9yoNOSwalRcKEnCP9zG6tVS0oivo5tCeuH6AaaW7Jpc,890
-ominfra/supervisor/supervisor.py,sha256=3yBaGJHJGqDXlARjpHWg7371N6jb29HaFEI-_fSp-xU,12028
-ominfra/supervisor/types.py,sha256=i2C7ZBvLqjVliIYAYQnPx2WwIQ14HIVV1hTUgB2mbBM,4740
+ominfra/supervisor/supervisor.py,sha256=bLl4cHu6G4Kr_nDAr6Q6jG6g5oRWIx0rQaPHGVG0WMQ,9359
+ominfra/supervisor/types.py,sha256=TZYvdX_3XQ82PgpOOJF2PT5VMlOD9zqMVZhB-ZofTRI,4891
 ominfra/supervisor/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ominfra/supervisor/utils/collections.py,sha256=vcfmVYS4QngMdtEI1DvdRIcubmy55Wj40NCzW27_rIY,1361
 ominfra/supervisor/utils/diag.py,sha256=ujz4gkW7p3wmbaKFM8Hz5eHEwpoUkbB8JeDvcHilCz0,705
@@ -102,9 +103,9 @@ ominfra/tailscale/api.py,sha256=C5-t_b6jZXUWcy5k8bXm7CFnk73pSdrlMOgGDeGVrpw,1370
 ominfra/tailscale/cli.py,sha256=DSGp4hn5xwOW-l_u_InKlSF6kIobxtUtVssf_73STs0,3567
 ominfra/tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ominfra/tools/listresources.py,sha256=4qVg5txsb10EHhvqXXeM6gJ2jx9LbroEnPydDv1uXs0,6176
-ominfra-0.0.0.dev128.dist-info/LICENSE,sha256=B_hVtavaA8zCYDW99DYdcpDLKz1n3BBRjZrcbv8uG8c,1451
-ominfra-0.0.0.dev128.dist-info/METADATA,sha256=JV6Hw2ZOpDub73wDD4p69Qq9F6XbbKlthV7CmH6G_-k,731
-ominfra-0.0.0.dev128.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-ominfra-0.0.0.dev128.dist-info/entry_points.txt,sha256=kgecQ2MgGrM9qK744BoKS3tMesaC3yjLnl9pa5CRczg,37
-ominfra-0.0.0.dev128.dist-info/top_level.txt,sha256=E-b2OHkk_AOBLXHYZQ2EOFKl-_6uOGd8EjeG-Zy6h_w,8
-ominfra-0.0.0.dev128.dist-info/RECORD,,
+ominfra-0.0.0.dev129.dist-info/LICENSE,sha256=B_hVtavaA8zCYDW99DYdcpDLKz1n3BBRjZrcbv8uG8c,1451
+ominfra-0.0.0.dev129.dist-info/METADATA,sha256=VsF9d4SW55_llwPjsmpCXtrLrw5L0r86L5DndQ7YOwo,731
+ominfra-0.0.0.dev129.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
+ominfra-0.0.0.dev129.dist-info/entry_points.txt,sha256=kgecQ2MgGrM9qK744BoKS3tMesaC3yjLnl9pa5CRczg,37
+ominfra-0.0.0.dev129.dist-info/top_level.txt,sha256=E-b2OHkk_AOBLXHYZQ2EOFKl-_6uOGd8EjeG-Zy6h_w,8
+ominfra-0.0.0.dev129.dist-info/RECORD,,

ominfra/supervisor/context.py

@@ -1,80 +0,0 @@
-# ruff: noqa: UP006 UP007
-import errno
-import os
-import typing as ta
-
-from omlish.lite.logs import log
-
-from .configs import ServerConfig
-from .poller import Poller
-from .states import SupervisorState
-from .types import ServerContext
-from .types import ServerEpoch
-from .utils.fs import mktempfile
-from .utils.ostypes import Pid
-from .utils.ostypes import Rc
-
-
-class ServerContextImpl(ServerContext):
-    def __init__(
-            self,
-            config: ServerConfig,
-            poller: Poller,
-            *,
-            epoch: ServerEpoch = ServerEpoch(0),
-    ) -> None:
-        super().__init__()
-
-        self._config = config
-        self._poller = poller
-        self._epoch = epoch
-
-        self._state: SupervisorState = SupervisorState.RUNNING
-
-    @property
-    def config(self) -> ServerConfig:
-        return self._config
-
-    @property
-    def epoch(self) -> ServerEpoch:
-        return self._epoch
-
-    @property
-    def first(self) -> bool:
-        return not self._epoch
-
-    @property
-    def state(self) -> SupervisorState:
-        return self._state
-
-    def set_state(self, state: SupervisorState) -> None:
-        self._state = state
-
-    #
-
-    def waitpid(self) -> ta.Tuple[ta.Optional[Pid], ta.Optional[Rc]]:
-        # Need pthread_sigmask here to avoid concurrent sigchld, but Python doesn't offer in Python < 3.4.  There is
-        # still a race condition here; we can get a sigchld while we're sitting in the waitpid call. However, AFAICT, if
-        # waitpid is interrupted by SIGCHLD, as long as we call waitpid again (which happens every so often during the
-        # normal course in the mainloop), we'll eventually reap the child that we tried to reap during the interrupted
-        # call. At least on Linux, this appears to be true, or at least stopping 50 processes at once never left zombies
-        # lying around.
-        try:
-            pid, sts = os.waitpid(-1, os.WNOHANG)
-        except OSError as exc:
-            code = exc.args[0]
-            if code not in (errno.ECHILD, errno.EINTR):
-                log.critical('waitpid error %r; a process may not be cleaned up properly', code)
-            if code == errno.EINTR:
-                log.debug('EINTR during reap')
-            pid, sts = None, None
-        return pid, sts  # type: ignore
-
-    def get_auto_child_log_name(self, name: str, identifier: str, channel: str) -> str:
-        prefix = f'{name}-{channel}---{identifier}-'
-        logfile = mktempfile(
-            suffix='.log',
-            prefix=prefix,
-            dir=self.config.child_logdir,
-        )
-        return logfile