fix health req id gen when httpserver worker num > 1

lightllm/server/req_id_generator.py

@@ -1,3 +1,6 @@
+import os
+import psutil
+import sys
 import time
 import requests
 import numpy as np
@@ -27,6 +30,44 @@ def __init__(self):
         self.current_id.arr[0] = 0
         self.current_id.arr[1] = 0
         self.lock = AtomicShmLock(f"{get_unique_server_name()}_req_id_gen_lock")
+        self._wait_all_workers_ready()
+        logger.info("ReqIDGenerator init finished")
+
+    def _wait_all_workers_ready(self):
+        from lightllm.utils.envs_utils import get_unique_server_name
+        from lightllm.server.core.objs.shm_array import ShmArray
+
+        _sync_shm = ShmArray(
+            f"{get_unique_server_name()}_httpworker_start_sync", (self.args.httpserver_workers,), dtype=np.int64
+        )
+        _sync_shm.create_shm()
+        # 等待所有 httpserver 的 worker 启动完成，防止重新初始化对应的请求id 对应的shm
+        try_count = 0
+        while len(_find_sibling_processes()) + 1 != self.args.httpserver_workers:
+            time.sleep(0.1)
+            try_count += 1
+            if try_count > 120:
+                logger.error("wait all httpserver workers start failed")
+                sys.exit(-1)
+            else:
+                continue
+
+        cur_p_id = os.getpid()
+        pids = _find_sibling_processes()
+        pids.append(cur_p_id)
+        assert len(pids) == self.args.httpserver_workers
+        pids = sorted(pids)
+        index = pids.index(cur_p_id)
+        _sync_shm.arr[index] = cur_p_id
+        try_count = 0
+        while not all(a == b for a, b in zip(pids, _sync_shm.arr)):
+            time.sleep(0.1)
+            try_count += 1
+            if try_count > 120:
+                logger.error("wait all httpserver workers start failed 1")
+                sys.exit(-1)
+            else:
+                continue
 
     def _check_and_set_new_id_range(self):
         need_update_range = self.current_id.arr[0] + MAX_BEST_OF >= self.current_id.arr[1]
@@ -66,3 +107,30 @@ def generate_id(self):
 
 def convert_sub_id_to_group_id(sub_req_id):
     return (sub_req_id // MAX_BEST_OF) * MAX_BEST_OF
+
+
+def _find_sibling_processes():
+    # 获取当前进程的 PID
+    current_pid = os.getpid()
+
+    # 获取当前进程的信息
+    current_process = psutil.Process(current_pid)
+
+    # 获取当前进程的父进程
+    parent_process = current_process.parent()
+
+    if parent_process is None:
+        logger.error("Current process has no parent.")
+        return []
+
+    # 查找兄弟进程
+    sibling_processes = []
+    for proc in psutil.process_iter(["pid", "name"]):
+        try:
+            # 检查是否是兄弟进程（同一父进程且不是当前进程）
+            if proc.pid != current_pid and proc.ppid() == parent_process.pid:
+                sibling_processes.append(proc)
+        except (psutil.NoSuchProcess, psutil.AccessDenied):
+            continue
+
+    return [proc.pid for proc in sibling_processes]

lightllm/utils/health_check.py

@@ -8,16 +8,12 @@
 from lightllm.server.httpserver.manager import HttpServerManager
 from lightllm.server.router.dynamic_prompt.shared_arr import SharedInt
 from fastapi import Request
-from lightllm.server.req_id_generator import ReqIDGenerator
 from lightllm.utils.log_utils import init_logger
 from lightllm.utils.envs_utils import get_unique_server_name, get_env_start_args
 
 logger = init_logger(__name__)
 
 
-_g_health_req_id_gen = ReqIDGenerator()
-
-
 @dataclass
 class HealthObj:
     _is_health: bool = False
@@ -81,9 +77,9 @@ async def health_check(args, httpserver_manager: HttpServerManager, request: Req
         if get_env_start_args().run_mode == "pd_master":
             # Since the id assigned by pd master needs to be passed to prefill and decode nodes for inference,
             # a normal request id is required instead of a negative id.
-            sampling_params.group_request_id = _g_health_req_id_gen.generate_id()
+            sampling_params.group_request_id = httpserver_manager.id_gen.generate_id()
         else:
-            sampling_params.group_request_id = -_g_health_req_id_gen.generate_id()  # health monitor 的 id 是负的
+            sampling_params.group_request_id = -httpserver_manager.id_gen.generate_id()  # health monitor 的 id 是负的
         multimodal_params_dict = request_dict.get("multimodal_params", {})
         multimodal_params = MultimodalParams(**multimodal_params_dict)
         results_generator = httpserver_manager.generate(