refactor(pymsg): 新增AsyncRpcQueue通用异步队列，log_client基于此重构
- pymsg新增AsyncRpcQueue：提供内存队列+后台线程+攒批处理的通用异步基础设施， 支持两种模式：简单RPC模式(client+method+serialize)和自定义handler模式 - confs/dnfs/static的log_client.py重构为基于AsyncRpcQueue的薄封装， 保持API完全向后兼容(push/info/error/warning/debug/critical + 全局logs单例) - 消除了三份log_client副本中重复的队列/线程/攒批逻辑(约150行/份)， 代码量净减少约50行 - LogClient的行为与旧版一致：逐条INSERT到db、fail-fast网络错误、队列满丢最旧
This commit is contained in:
qing committed
1 parent
101aed48be
commit
d883dc620e
5 files changed
+384
-433
No files matched your search
@@ -0,0 +1,202 @@
|
||||
"""
|
||||
AsyncRpcQueue - 通用异步批量发送队列
|
||||
|
||||
提供内存队列 + 后台线程的异步批量处理机制，适用于日志上报、指标采集等
|
||||
非关键路径场景。业务线程put操作非阻塞，后台线程攒批处理，失败时通过错误回调
|
||||
通知，不影响主业务流程。
|
||||
|
||||
两种使用模式：
|
||||
|
||||
1. 简单RPC模式（指定client + method + serialize）：
|
||||
queue = AsyncRpcQueue(
|
||||
client=rpc_client,
|
||||
method='log_api.push',
|
||||
batch_size=50,
|
||||
flush_interval=2.0,
|
||||
serialize=lambda items: {'logs': json.dumps(items)},
|
||||
)
|
||||
|
||||
2. 自定义处理模式（指定handler回调）：
|
||||
def send_batch(batch):
|
||||
for item in batch:
|
||||
rpc_client.call('execute', item)
|
||||
queue = AsyncRpcQueue(
|
||||
handler=send_batch,
|
||||
batch_size=50,
|
||||
flush_interval=2.0,
|
||||
)
|
||||
"""
|
||||
|
||||
import threading
|
||||
import time
|
||||
from collections import deque
|
||||
from typing import Any, Callable, Optional
|
||||
|
||||
|
||||
class AsyncRpcQueue:
|
||||
"""通用异步批量发送队列"""
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
client: Any = None,
|
||||
method: str = None,
|
||||
*,
|
||||
handler: Optional[Callable[[list], None]] = None,
|
||||
batch_size: int = 100,
|
||||
flush_interval: float = 2.0,
|
||||
max_queue_size: int = 10000,
|
||||
serialize: Optional[Callable[[list], dict]] = None,
|
||||
on_error: Optional[Callable[[Exception, list], None]] = None,
|
||||
retry_on_refused: bool = False,
|
||||
retry_interval: float = 1.0,
|
||||
):
|
||||
"""
|
||||
初始化异步队列
|
||||
|
||||
Args:
|
||||
client: RpcClient实例（RPC模式必填）
|
||||
method: RPC方法名（RPC模式必填）
|
||||
handler: 自定义批量处理回调，接收一批items列表。提供此参数时
|
||||
忽略client/method/serialize/retry_on_refused。
|
||||
batch_size: 批量发送的阈值，达到此数量立即发送
|
||||
flush_interval: 定时flush间隔（秒）
|
||||
max_queue_size: 队列最大长度，超过则丢弃最旧数据
|
||||
serialize: 序列化函数，接收items列表返回RPC参数字典
|
||||
默认: lambda items: {'items': items}
|
||||
on_error: 错误回调，接收(exception, batch_items)
|
||||
retry_on_refused: 是否在连接被拒绝时重试（RPC模式，透传给client.call）
|
||||
retry_interval: 重试间隔（秒）（RPC模式，透传给client.call）
|
||||
"""
|
||||
if handler is None:
|
||||
if client is None or method is None:
|
||||
raise ValueError("必须提供 handler 或 (client + method)")
|
||||
self._handler = self._make_rpc_handler(
|
||||
client, method,
|
||||
serialize or (lambda items: {'items': items}),
|
||||
retry_on_refused, retry_interval,
|
||||
)
|
||||
else:
|
||||
self._handler = handler
|
||||
|
||||
self._batch_size = batch_size
|
||||
self._flush_interval = flush_interval
|
||||
self._max_queue_size = max_queue_size
|
||||
self._on_error = on_error
|
||||
|
||||
self._queue: deque = deque()
|
||||
self._lock = threading.Lock()
|
||||
self._wakeup = threading.Event()
|
||||
self._thread: Optional[threading.Thread] = None
|
||||
self._closed = False
|
||||
self._dropped_count = 0
|
||||
|
||||
@staticmethod
|
||||
def _make_rpc_handler(client, method, serialize, retry_on_refused, retry_interval):
|
||||
"""创建RPC模式的默认handler"""
|
||||
def handler(batch: list):
|
||||
params = serialize(batch)
|
||||
client.call(
|
||||
method, params,
|
||||
retry_on_refused=retry_on_refused,
|
||||
retry_interval=retry_interval,
|
||||
)
|
||||
return handler
|
||||
|
||||
def start(self) -> None:
|
||||
"""启动后台发送线程"""
|
||||
if self._thread is not None and self._thread.is_alive():
|
||||
return
|
||||
self._closed = False
|
||||
self._thread = threading.Thread(
|
||||
target=self._run,
|
||||
name=f'AsyncRpcQueue',
|
||||
daemon=True,
|
||||
)
|
||||
self._thread.start()
|
||||
|
||||
def put(self, item: Any) -> bool:
|
||||
"""
|
||||
放入一条数据到队列（非阻塞）
|
||||
|
||||
Args:
|
||||
item: 要发送的数据项
|
||||
|
||||
Returns:
|
||||
bool: True表示放入成功，False表示队列已关闭
|
||||
"""
|
||||
if self._closed:
|
||||
return False
|
||||
with self._lock:
|
||||
if len(self._queue) >= self._max_queue_size:
|
||||
self._queue.popleft()
|
||||
self._dropped_count += 1
|
||||
self._queue.append(item)
|
||||
self._wakeup.set()
|
||||
return True
|
||||
|
||||
def flush(self) -> None:
|
||||
"""立即触发一次发送（会阻塞直到处理完成，主要用于关闭时）"""
|
||||
self._send_batch()
|
||||
|
||||
def close(self) -> None:
|
||||
"""关闭队列，flush剩余数据后停止后台线程"""
|
||||
if self._closed:
|
||||
return
|
||||
self._closed = True
|
||||
self._wakeup.set()
|
||||
if self._thread is not None:
|
||||
self._thread.join(timeout=5.0)
|
||||
self._send_batch()
|
||||
|
||||
@property
|
||||
def pending_count(self) -> int:
|
||||
"""当前队列中待发送的数据量"""
|
||||
with self._lock:
|
||||
return len(self._queue)
|
||||
|
||||
@property
|
||||
def dropped_count(self) -> int:
|
||||
"""因队列满而丢弃的数据量"""
|
||||
return self._dropped_count
|
||||
|
||||
def _run(self) -> None:
|
||||
"""后台线程主循环"""
|
||||
last_flush = time.time()
|
||||
while not self._closed:
|
||||
self._wakeup.wait(timeout=0.1)
|
||||
self._wakeup.clear()
|
||||
|
||||
now = time.time()
|
||||
should_flush = False
|
||||
with self._lock:
|
||||
if len(self._queue) >= self._batch_size:
|
||||
should_flush = True
|
||||
elif now - last_flush >= self._flush_interval and self._queue:
|
||||
should_flush = True
|
||||
if should_flush:
|
||||
self._send_batch()
|
||||
last_flush = time.time()
|
||||
|
||||
def _send_batch(self) -> None:
|
||||
"""取一批数据处理"""
|
||||
with self._lock:
|
||||
if not self._queue:
|
||||
return
|
||||
batch = list(self._queue)
|
||||
self._queue.clear()
|
||||
try:
|
||||
self._handler(batch)
|
||||
except Exception as e:
|
||||
if self._on_error:
|
||||
try:
|
||||
self._on_error(e, batch)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
def __enter__(self):
|
||||
self.start()
|
||||
return self
|
||||
|
||||
def __exit__(self, exc_type, exc_val, exc_tb):
|
||||
self.close()
|
||||
return False
|
||||
Reference in new issue
Block a user