Use a separate thread to watch for dead processes

taskiq/cli/worker/process_manager.py

@@ -2,8 +2,9 @@
 import signal
 from dataclasses import dataclass
 from multiprocessing import Process, Queue
+from threading import Thread
 from time import sleep
-from typing import Any, Callable, List
+from typing import Any, Callable, List, NoReturn
 
 from watchdog.observers import Observer
 
@@ -172,7 +173,32 @@ def prepare_workers(self) -> None:
             )
             self.workers.append(work_proc)
 
-    def start(self) -> None:  # noqa: C901, WPS213
+    def start_worker_watcher(self) -> None:
+        """
+        Start worker watcher thread.
+
+        The worker watcher thread periodically checks
+        that the worker processes are alive and schedules
+        a reload for ones that failed.
+        """
+
+        def watcher(
+            workers: List[Process],
+            action_queue: Queue[ProcessActionBase],
+        ) -> NoReturn:
+            logging.info("Started the worker watcher thread.")
+            while True:  # noqa: WPS457
+                sleep(1.0)
+                for worker_num, worker in enumerate(workers):
+                    if not worker.is_alive():
+                        logger.info(f"{worker.name} is dead. Scheduling reload.")
+                        action_queue.put(ReloadOneAction(worker_num=worker_num))
+
+        self._watcher = Thread(target=watcher, args=(self.workers, self.action_queue))
+        self._watcher.daemon = True
+        self._watcher.start()
+
+    def start(self) -> None:
         """
         Start managing child processes.
 
@@ -199,29 +225,17 @@ def start(self) -> None:  # noqa: C901, WPS213
         some reason, it schedules a restart for dead process.
         """
         self.prepare_workers()
+        self.start_worker_watcher()
         while True:
-            sleep(1)
-            reloaded_workers = set()
-            # We bulk_process all pending events.
-            while not self.action_queue.empty():
-                action = self.action_queue.get()
-                logging.debug(f"Got event: {action}")
-                if isinstance(action, ReloadAllAction):
-                    action.handle(
-                        workers_num=len(self.workers),
-                        action_queue=self.action_queue,
-                    )
-                elif isinstance(action, ReloadOneAction):
-                    # If we just reloaded this worker, skip handling.
-                    if action.worker_num in reloaded_workers:
-                        continue
-                    action.handle(self.workers, self.args, self.worker_function)
-                    reloaded_workers.add(action.worker_num)
-                elif isinstance(action, ShutdownAction):
-                    logger.debug("Process manager closed.")
-                    return
-
-            for worker_num, worker in enumerate(self.workers):
-                if not worker.is_alive():
-                    logger.info(f"{worker.name} is dead. Scheduling reload.")
-                    self.action_queue.put(ReloadOneAction(worker_num=worker_num))
+            action = self.action_queue.get()
+            logging.debug(f"Got event: {action}")
+            if isinstance(action, ReloadAllAction):
+                action.handle(
+                    workers_num=len(self.workers),
+                    action_queue=self.action_queue,
+                )
+            elif isinstance(action, ReloadOneAction):
+                action.handle(self.workers, self.args, self.worker_function)
+            elif isinstance(action, ShutdownAction):
+                logger.debug("Process manager closed.")
+                return