Merge pull request #112 from wangjie5540/dev/wtg

optimize logging for network issues.

metagpt/config.py

@@ -46,7 +46,6 @@ class Config(metaclass=Singleton):
         self.openai_api_key = self._get("OPENAI_API_KEY")
         if not self.openai_api_key or "YOUR_API_KEY" == self.openai_api_key:
             raise NotConfiguredException("Set OPENAI_API_KEY first")
-
         self.openai_api_base = self._get("OPENAI_API_BASE")
         if not self.openai_api_base or "YOUR_API_BASE" == self.openai_api_base:
             openai_proxy = self._get("OPENAI_PROXY") or self.global_proxy
@@ -67,11 +66,11 @@ class Config(metaclass=Singleton):
         self.google_api_key = self._get("GOOGLE_API_KEY")
         self.google_cse_id = self._get("GOOGLE_CSE_ID")
         self.search_engine = self._get("SEARCH_ENGINE", SearchEngineType.SERPAPI_GOOGLE)
- 
+
         self.web_browser_engine = WebBrowserEngineType(self._get("WEB_BROWSER_ENGINE", "playwright"))
         self.playwright_browser_type = self._get("PLAYWRIGHT_BROWSER_TYPE", "chromium")
         self.selenium_browser_type = self._get("SELENIUM_BROWSER_TYPE", "chrome")
-      
+
         self.long_term_memory = self._get('LONG_TERM_MEMORY', False)
         if self.long_term_memory:
             logger.warning("LONG_TERM_MEMORY is True")

metagpt/provider/openai_api.py

@@ -6,10 +6,11 @@
 """
 import asyncio
 import time
-from functools import wraps
 from typing import NamedTuple
 
 import openai
+from openai.error import APIConnectionError
+from tenacity import retry, stop_after_attempt, after_log, wait_fixed, retry_if_exception_type
 
 from metagpt.config import CONFIG
 from metagpt.logs import logger
@@ -23,23 +24,6 @@ from metagpt.utils.token_counter import (
 )
 
 
-def retry(max_retries):
-    def decorator(f):
-        @wraps(f)
-        async def wrapper(*args, **kwargs):
-            for i in range(max_retries):
-                try:
-                    return await f(*args, **kwargs)
-                except Exception:
-                    if i == max_retries - 1:
-                        raise
-                    await asyncio.sleep(2**i)
-
-        return wrapper
-
-    return decorator
-
-
 class RateLimiter:
     """Rate control class, each call goes through wait_if_needed, sleep if rate control is needed"""
 
@@ -132,6 +116,15 @@ class CostManager(metaclass=Singleton):
         return Costs(self.total_prompt_tokens, self.total_completion_tokens, self.total_cost, self.total_budget)
 
 
+def log_and_reraise(retry_state):
+    logger.error(f"Retry attempts exhausted. Last exception: {retry_state.outcome.exception()}")
+    logger.warning("""
+Recommend going to https://deepwisdom.feishu.cn/wiki/MsGnwQBjiif9c3koSJNcYaoSnu4#part-XdatdVlhEojeAfxaaEZcMV3ZniQ
+See FAQ 5.8
+""")
+    raise retry_state.outcome.exception()
+
+
 class OpenAIGPTAPI(BaseGPTAPI, RateLimiter):
     """
     Check https://platform.openai.com/examples for examples
@@ -193,6 +186,7 @@ class OpenAIGPTAPI(BaseGPTAPI, RateLimiter):
                 "stop": None,
                 "temperature": 0.3,
             }
+        kwargs["timeout"] = 3
         return kwargs
 
     async def _achat_completion(self, messages: list[dict]) -> dict:
@@ -215,7 +209,13 @@ class OpenAIGPTAPI(BaseGPTAPI, RateLimiter):
         #     messages = self.messages_to_dict(messages)
         return await self._achat_completion(messages)
 
-    @retry(max_retries=6)
+    @retry(
+        stop=stop_after_attempt(3),
+        wait=wait_fixed(1),
+        after=after_log(logger, logger.level('WARNING').name),
+        retry=retry_if_exception_type(APIConnectionError),
+        retry_error_callback=log_and_reraise,
+    )
     async def acompletion_text(self, messages: list[dict], stream=False) -> str:
         """when streaming, print each token in place."""
         if stream: