jettask 0.2.18__py3-none-any.whl → 0.2.20__py3-none-any.whl

jettask/executors/base.py

@@ -1,30 +0,0 @@
-from collections import deque
-from abc import ABC, abstractmethod
-import time
-from typing import TYPE_CHECKING
-
-from .common import CommonExecutorMixin
-
-if TYPE_CHECKING:
-    from ..core.app import Jettask
-
-
-class BaseExecutor(CommonExecutorMixin, ABC):
-    """Base class for all executors"""
-    
-    def __init__(self, event_queue: deque, app: "Jettask", concurrency: int = 1) -> None:
-        self.event_queue = event_queue
-        self.app = app
-        self.concurrency = concurrency
-        self.last_refresh_pending_time = 0
-        self.pedding_count = 0
-        self.batch_counter = 0
-
-    def logic(self, *args, **kwargs):
-        """Process a single task"""
-        pass
-
-    @abstractmethod
-    def loop(self):
-        """Main loop for the executor"""
-        pass

jettask/executors/common.py

@@ -1,148 +0,0 @@
-import time
-import traceback
-from ..utils.serializer import dumps_str, loads_str
-from typing import TYPE_CHECKING, Any, Dict, Optional, Tuple
-from ..utils.traceback_filter import filter_framework_traceback
-
-if TYPE_CHECKING:
-    from ..core.app import Jettask
-    from ..core.task import Task
-
-
-class CommonExecutorMixin:
-    """Mixin class containing common functionality for all executors"""
-    
-    app: "Jettask"
-    last_refresh_pending_time: float
-    pedding_count: int
-    
-    def parse_task_data(self, data: bytes) -> Tuple[str, str, str, list, dict]:
-        """Parse task data from bytes"""
-        data_str = data.decode("utf-8")
-        parts = data_str.split("_____", 4)
-        event_id = parts[0]
-        task_name = parts[1]
-        trigger_time = parts[2]
-        args = loads_str(parts[3]) if parts[3] else []
-        kwargs = loads_str(parts[4]) if parts[4] else {}
-        return event_id, task_name, trigger_time, args, kwargs
-    
-    def get_task(self, task_name: str) -> Optional["Task"]:
-        """Get task by name"""
-        return self.app.get_task_by_name(task_name)
-    
-    def handle_task_error(self, event_id: str, task_name: str, error: Exception) -> Dict[str, Any]:
-        """Handle task execution error"""
-        # 使用过滤后的堆栈信息
-        error_msg = filter_framework_traceback()
-        result = {
-            "status": "error",
-            "task_name": task_name,
-            "message": str(error),
-            "traceback": error_msg
-        }
-        self.app.set_task_status(event_id, "error")
-        self.app.set_data(event_id, dumps_str(result))
-        return result
-    
-    def handle_task_success(self, event_id: str, task_name: str, result: Any) -> Dict[str, Any]:
-        """Handle successful task execution"""
-        result_data = {
-            "status": "success",
-            "task_name": task_name,
-            "result": result
-        }
-        self.app.set_task_status(event_id, "success")
-        self.app.set_data(event_id, dumps_str(result_data))
-        return result_data
-    
-    def ack_message(self, data: bytes) -> None:
-        """Acknowledge message processing"""
-        self.app.ep.ack(data)
-    
-    def get_routing_data(self, kwargs: dict) -> Tuple[Optional[str], Optional[str]]:
-        """Extract routing data from kwargs"""
-        routing_key = kwargs.pop("routing_key", None)
-        aggregation_key = kwargs.pop("aggregation_key", None)
-        return routing_key, aggregation_key
-    
-    def should_process_routing(self, routing_key: Optional[str], aggregation_key: Optional[str]) -> bool:
-        """Check if task should be processed based on routing"""
-        if routing_key:
-            # Check solo running state
-            if self.app.is_solo_running_by_aggregation_key(aggregation_key):
-                return False
-            # Set solo running state
-            self.app.set_solo_running_by_aggregation_key(aggregation_key)
-        return True
-    
-    def clear_routing_state(self, routing_key: Optional[str], aggregation_key: Optional[str]) -> None:
-        """Clear routing state after task completion"""
-        if routing_key:
-            self.app.clear_solo_running_by_aggregation_key(aggregation_key)
-    
-    def handle_urgent_retry(self, kwargs: dict, event_id: str, task_name: str, 
-                          trigger_time: str, args: list) -> bool:
-        """Handle urgent retry logic"""
-        urgent_retry = kwargs.pop("urgent_retry", None)
-        if urgent_retry:
-            delay = kwargs.pop("delay", 0)
-            self.app.delay(
-                task_name=task_name,
-                delay=delay,
-                args=args,
-                kwargs=kwargs,
-                task_id=event_id,
-                trigger_time=trigger_time,
-                producer_type="urgent_retry",
-            )
-            self.app.set_task_status(event_id, "retry")
-            return True
-        return False
-    
-    def get_pending_count(self) -> int:
-        """Get pending count with caching"""
-        current_time = time.time()
-        if current_time - self.last_refresh_pending_time > 1:
-            self.pedding_count = self.app.ep.get_pending_count()
-            self.last_refresh_pending_time = current_time
-        return self.pedding_count
-    
-    def execute_task_lifecycle(self, task: "Task", event_id: str, trigger_time: str,
-                             args: list, kwargs: dict, is_async: bool = False):
-        """Execute task with lifecycle methods"""
-        # This method will be implemented differently for sync/async executors
-        raise NotImplementedError("Subclasses must implement execute_task_lifecycle")
-    
-    def format_batch_data(self, data: bytes) -> Tuple[str, str, list, dict]:
-        """Parse batch task data"""
-        data_str = data.decode("utf-8")
-        parts = data_str.split("_____", 3)
-        batch_id = parts[0]
-        task_name = parts[1]
-        args = loads_str(parts[2]) if parts[2] else []
-        kwargs = loads_str(parts[3]) if parts[3] else {}
-        return batch_id, task_name, args, kwargs
-    
-    def extract_batch_params(self, kwargs: dict) -> Tuple[list, list, str, Optional[str], Optional[str]]:
-        """Extract batch-specific parameters from kwargs"""
-        event_ids = kwargs.pop("event_ids", [])
-        trigger_times = kwargs.pop("trigger_times", [])
-        producer_type = kwargs.pop("producer_type", "normal")
-        routing_key = kwargs.pop("routing_key", None)
-        aggregation_key = kwargs.pop("aggregation_key", None)
-        return event_ids, trigger_times, producer_type, routing_key, aggregation_key
-    
-    def ack_batch_events(self, event_ids: list) -> None:
-        """Acknowledge multiple events in batch"""
-        for event_id in event_ids:
-            data = f"{event_id}_____batch_ack".encode()
-            self.ack_message(data)
-    
-    def update_batch_status(self, event_ids: list, status: str, result: Any = None) -> None:
-        """Update status for multiple events in batch"""
-        result_str = dumps_str(result) if result is not None else None
-        for event_id in event_ids:
-            self.app.set_task_status(event_id, status)
-            if result_str:
-                self.app.set_data(event_id, result_str)

jettask/executors/multi_asyncio.py

@@ -1,309 +0,0 @@
-import multiprocessing
-import logging
-import time
-import os
-import signal
-from typing import Dict, List
-from multiprocessing import Process, Event
-from dataclasses import dataclass
-
-from .base import BaseExecutor
-
-logger = logging.getLogger('app')
-
-
-@dataclass
-class ProcessConfig:
-    """Configuration for AsyncIO executor process"""
-    executor_id: int
-    redis_url: str
-    queues: List[str]
-    app_tasks: Dict
-    consumer_strategy: str
-    consumer_config: Dict
-    max_connections: int
-    prefetch_multiplier: int
-    concurrency_per_process: int = 10000
-
-
-class MultiAsyncioExecutor(BaseExecutor):
-    """
-    Multi-asyncio executor that manages multiple AsyncioExecutor instances
-    Each instance runs in its own process with its own event loop
-    
-    Features:
-    - Automatic process restart on failure
-    - Graceful shutdown with timeout
-    - Process health monitoring
-    - Configurable concurrency per process
-    """
-    
-    def __init__(self, event_queue, app, concurrency=3):
-        super().__init__(event_queue, app, concurrency)
-        self.processes: Dict[int, Process] = {}
-        self.process_configs: Dict[int, ProcessConfig] = {}
-        self.shutdown_event = Event()
-        self._monitor_interval = 1  # seconds
-        self._status_log_interval = 30  # seconds
-        self._restart_delay = 2  # seconds
-        self._max_restart_attempts = 3
-        self._restart_counts: Dict[int, int] = {}
-        self._main_received_signal = False  # Track if main process received signal
-        
-    def logic(self):
-        """
-        Logic method for BaseExecutor interface.
-        Not used in MultiAsyncioExecutor as it delegates to AsyncioExecutor instances.
-        """
-        pass
-        
-    @staticmethod
-    def _run_asyncio_executor(config: ProcessConfig, shutdown_event):
-        """Run an AsyncioExecutor in a separate process"""
-        # Set process name for debugging
-        multiprocessing.current_process().name = f"AsyncioExecutor-{config.executor_id}"
-        
-        # Configure logging for subprocess
-        logging.basicConfig(
-            level=logging.INFO,
-            format=f"%(asctime)s - %(levelname)s - [Executor-{config.executor_id}] - %(message)s",
-            datefmt="%Y-%m-%d %H:%M:%S",
-        )
-        logger = logging.getLogger('app')
-        
-        # Handle signals gracefully
-        def signal_handler(signum, _frame):
-            logger.info(f"AsyncioExecutor #{config.executor_id} received signal {signum}")
-            shutdown_event.set()
-            
-        signal.signal(signal.SIGINT, signal_handler)
-        signal.signal(signal.SIGTERM, signal_handler)
-        
-        try:
-            # Import inside process to avoid pickle issues
-            from ..core.app import Jettask
-            
-            # Create app instance for this process
-            app = Jettask(
-                redis_url=config.redis_url,
-                max_connections=config.max_connections,
-                consumer_strategy=config.consumer_strategy,
-                consumer_config=config.consumer_config,
-                tasks=config.app_tasks
-            )
-            
-            logger.info(f"Started AsyncioExecutor #{config.executor_id} in process {os.getpid()}")
-            
-            # Check shutdown event before starting
-            if shutdown_event.is_set():
-                logger.info(f"AsyncioExecutor #{config.executor_id} shutdown before start")
-                return
-            
-            # Start the executor
-            app._start(
-                execute_type="asyncio",
-                queues=config.queues,
-                concurrency=config.concurrency_per_process,
-                prefetch_multiplier=config.prefetch_multiplier
-            )
-            
-        except KeyboardInterrupt:
-            logger.info(f"AsyncioExecutor #{config.executor_id} received interrupt")
-        except Exception as e:
-            logger.error(f"AsyncioExecutor #{config.executor_id} error: {e}")
-            import traceback
-            traceback.print_exc()
-            raise  # Re-raise to trigger restart mechanism
-        finally:
-            logger.info(f"AsyncioExecutor #{config.executor_id} stopped")
-    
-    def _create_process_config(self, executor_id: int, queues: List[str], 
-                              prefetch_multiplier: int) -> ProcessConfig:
-        """Create configuration for a process"""
-        return ProcessConfig(
-            executor_id=executor_id,
-            redis_url=self.app.redis_url,
-            queues=queues,
-            app_tasks=self.app._tasks,
-            consumer_strategy=self.app.consumer_strategy,
-            consumer_config=self.app.consumer_config,
-            max_connections=self.app.max_connections,
-            prefetch_multiplier=prefetch_multiplier,
-            concurrency_per_process=10000
-        )
-    
-    def _start_process(self, executor_id: int, config: ProcessConfig) -> Process:
-        """Start a single AsyncioExecutor process"""
-        process = Process(
-            target=self._run_asyncio_executor,
-            args=(config, self.shutdown_event),
-            name=f"AsyncioExecutor-{executor_id}"
-        )
-        process.start()
-        logger.info(f"Started AsyncioExecutor process #{executor_id} (PID: {process.pid})")
-        return process
-    
-    def _restart_process(self, executor_id: int):
-        """Restart a failed process with exponential backoff"""
-        if self.shutdown_event.is_set():
-            return
-            
-        restart_count = self._restart_counts.get(executor_id, 0)
-        if restart_count >= self._max_restart_attempts:
-            logger.error(f"Process #{executor_id} exceeded max restart attempts ({self._max_restart_attempts})")
-            return
-            
-        self._restart_counts[executor_id] = restart_count + 1
-        delay = self._restart_delay * (2 ** restart_count)  # Exponential backoff
-        
-        logger.info(f"Restarting process #{executor_id} (attempt {restart_count + 1}/{self._max_restart_attempts}) "
-                   f"after {delay}s delay")
-        time.sleep(delay)
-        
-        config = self.process_configs[executor_id]
-        process = self._start_process(executor_id, config)
-        self.processes[executor_id] = process
-    
-    def _monitor_processes(self):
-        """Monitor process health and restart failed processes"""
-        alive_count = 0
-        for executor_id, process in list(self.processes.items()):
-            if process.is_alive():
-                alive_count += 1
-                # Reset restart count for healthy processes
-                self._restart_counts[executor_id] = 0
-            else:
-                exit_code = process.exitcode
-                # 如果正在关闭，不要重启进程
-                if self.shutdown_event.is_set():
-                    logger.info(f"Process #{executor_id} stopped during shutdown")
-                # 如果进程收到SIGTERM（exit_code == -15）或SIGINT（exit_code == -2），认为是正常关闭
-                elif exit_code == -15 or exit_code == -2:
-                    logger.info(f"Process #{executor_id} received termination signal, marking as shutdown")
-                    # 设置shutdown事件，避免重启其他进程
-                    self.shutdown_event.set()
-                # 如果所有进程都同时停止（正常退出），认为是收到了关闭信号
-                elif exit_code == 0 or exit_code is None:
-                    # 如果主进程已经收到信号，不要重启
-                    if self._main_received_signal:
-                        logger.info(f"Process #{executor_id} stopped after main received signal")
-                    else:
-                        # 检查是否所有进程都停止了
-                        all_stopped = True
-                        for _, p in self.processes.items():
-                            if p.is_alive():
-                                all_stopped = False
-                                break
-                        
-                        if all_stopped:
-                            logger.info(f"All processes stopped simultaneously, marking as shutdown")
-                            self.shutdown_event.set()
-                        else:
-                            logger.warning(f"Process #{executor_id} stopped unexpectedly with exit code {exit_code}")
-                            self._restart_process(executor_id)
-                else:
-                    logger.warning(f"Process #{executor_id} exited with code {exit_code}")
-                    self._restart_process(executor_id)
-        
-        return alive_count
-    
-    def loop(self):
-        """Main loop that starts and monitors AsyncioExecutor processes"""
-        logger.info(f"Starting MultiAsyncioExecutor with {self.concurrency} asyncio processes")
-        
-        # Set up signal handler to track when main process receives signal
-        def signal_handler(signum, frame):
-            logger.info(f"MultiAsyncioExecutor received signal {signum}")
-            self._main_received_signal = True
-            self.shutdown_event.set()
-        
-        # Register signal handlers
-        signal.signal(signal.SIGINT, signal_handler)
-        signal.signal(signal.SIGTERM, signal_handler)
-        
-        try:
-            # Get configuration
-            queues = getattr(self.app.ep, 'queues', ['robust_bench'])
-            prefetch_multiplier = getattr(self, 'prefetch_multiplier', 100)
-            
-            # Start AsyncioExecutor processes
-            for i in range(self.concurrency):
-                config = self._create_process_config(i, queues, prefetch_multiplier)
-                self.process_configs[i] = config
-                
-                process = self._start_process(i, config)
-                self.processes[i] = process
-                
-                # Small delay to avoid thundering herd
-                time.sleep(0.1)
-            
-            logger.info(f"All {self.concurrency} AsyncioExecutor processes started")
-            
-            # Monitor executor processes
-            last_status_log = time.time()
-            
-            while not self.shutdown_event.is_set():
-                # Monitor process health
-                alive_count = self._monitor_processes()
-                
-                if alive_count == 0:
-                    if self._main_received_signal or self.shutdown_event.is_set():
-                        logger.info("All AsyncioExecutor processes have stopped during shutdown")
-                    else:
-                        logger.error("All AsyncioExecutor processes have stopped unexpectedly")
-                    break
-                
-                # Log status periodically
-                current_time = time.time()
-                if current_time - last_status_log > self._status_log_interval:
-                    dead_count = self.concurrency - alive_count
-                    logger.info(f"MultiAsyncioExecutor status - Active: {alive_count}/{self.concurrency}, "
-                              f"Dead: {dead_count}, Restart attempts: {sum(self._restart_counts.values())}")
-                    last_status_log = current_time
-                
-                time.sleep(self._monitor_interval)
-                
-        except KeyboardInterrupt:
-            logger.info("MultiAsyncioExecutor received KeyboardInterrupt")
-            self._main_received_signal = True
-            self.shutdown_event.set()
-        except Exception as e:
-            logger.error(f"MultiAsyncioExecutor error: {e}")
-            import traceback
-            traceback.print_exc()
-        finally:
-            self.shutdown()
-    
-    def shutdown(self):
-        """Gracefully shutdown all executor processes"""
-        logger.info("Shutting down MultiAsyncioExecutor...")
-        
-        # Signal shutdown - this prevents any restart attempts
-        self.shutdown_event.set()
-        
-        # 快速关闭模式 - 减少等待时间
-        shutdown_timeout = 1.0  # 减少到1秒
-        terminate_timeout = 0.5  # 减少到0.5秒
-        
-        # 先发送终止信号给所有进程
-        for executor_id, process in self.processes.items():
-            if process.is_alive():
-                logger.info(f"Sending TERM signal to process #{executor_id} (PID: {process.pid})")
-                process.terminate()
-        
-        # 短暂等待所有进程自行退出
-        time.sleep(shutdown_timeout)
-        
-        # 强制杀死仍在运行的进程
-        for executor_id, process in self.processes.items():
-            if process.is_alive():
-                logger.warning(f"Process #{executor_id} did not terminate, killing...")
-                process.kill()
-                # 不再等待join，让操作系统清理
-        
-        # Clear process tracking
-        self.processes.clear()
-        self.process_configs.clear()
-        self._restart_counts.clear()
-        
-        logger.info("MultiAsyncioExecutor shutdown complete")