Skip to content

Speakr is a personal, self-hosted web application designed for transcribing audio recordings

Python

Always log resolved LLM timeout config and surface elapsed time on APITimeoutError

Murtaza Nasir7293210b4264
diff --git a/src/services/llm.py b/src/services/llm.py
index 9da66ffb..2ee27cb3 100644
--- a/src/services/llm.py
+++ b/src/services/llm.py
@@ -5,9 +5,10 @@ LLM API integration services (OpenAI/OpenRouter).
import os
import re
import json
+import time
import logging
import httpx
-from openai import OpenAI
+from openai import OpenAI, APITimeoutError
# Use standard logging instead of current_app.logger for context independence
logger = logging.getLogger(__name__)
@@ -48,7 +49,14 @@ ENABLE_STREAM_OPTIONS = os.environ.get("ENABLE_STREAM_OPTIONS", "true").lower()
# Connect/write stay short so misconfigured URLs fail fast instead of hanging
LLM_REQUEST_TIMEOUT = int(os.environ.get("LLM_REQUEST_TIMEOUT", "600"))
LLM_MAX_RETRIES = int(os.environ.get("LLM_MAX_RETRIES", "2"))
-llm_timeout = httpx.Timeout(connect=30.0, read=float(LLM_REQUEST_TIMEOUT), write=30.0, pool=30.0)
+LLM_CONNECT_TIMEOUT = float(os.environ.get("LLM_CONNECT_TIMEOUT", "30"))
+LLM_WRITE_TIMEOUT = float(os.environ.get("LLM_WRITE_TIMEOUT", "30"))
+llm_timeout = httpx.Timeout(
+ connect=LLM_CONNECT_TIMEOUT,
+ read=float(LLM_REQUEST_TIMEOUT),
+ write=LLM_WRITE_TIMEOUT,
+ pool=30.0,
+)
def get_chat_config():
@@ -96,8 +104,13 @@ try:
timeout=llm_timeout,
max_retries=LLM_MAX_RETRIES,
)
- if LLM_REQUEST_TIMEOUT != 600 or LLM_MAX_RETRIES != 2:
- logger.info(f"LLM client timeout: {LLM_REQUEST_TIMEOUT}s, max retries: {LLM_MAX_RETRIES}")
+ # Always log the resolved timeout/retry values at startup so users can confirm
+ # their LLM_REQUEST_TIMEOUT / LLM_MAX_RETRIES env vars actually took effect.
+ logger.info(
+ f"LLM client configured: read_timeout={LLM_REQUEST_TIMEOUT}s, "
+ f"connect_timeout={LLM_CONNECT_TIMEOUT}s, write_timeout={LLM_WRITE_TIMEOUT}s, "
+ f"max_retries={LLM_MAX_RETRIES}"
+ )
except Exception as client_init_e:
client = None
@@ -231,6 +244,7 @@ def call_llm_completion(messages, temperature=0.7, response_format=None, stream=
if response_format:
completion_args["response_format"] = response_format
+ request_started_at = time.monotonic()
response = client.chat.completions.create(**completion_args)
# Track usage for non-streaming calls
@@ -264,6 +278,17 @@ def call_llm_completion(messages, temperature=0.7, response_format=None, stream=
except TokenBudgetExceeded:
raise
+ except APITimeoutError as e:
+ elapsed = time.monotonic() - request_started_at if 'request_started_at' in locals() else None
+ elapsed_str = f"{elapsed:.1f}s" if elapsed is not None else "unknown"
+ logger.error(
+ f"LLM request timed out after {elapsed_str} (configured read_timeout={LLM_REQUEST_TIMEOUT}s, "
+ f"max_retries={LLM_MAX_RETRIES}, model={TEXT_MODEL_NAME}). "
+ f"If the elapsed time is much shorter than the configured timeout, an upstream proxy or the "
+ f"provider is closing the connection early. For reasoning models that take longer to think, "
+ f"increase LLM_REQUEST_TIMEOUT."
+ )
+ raise
except Exception as e:
logger.error(f"LLM API call failed: {e}")
raise
@@ -345,6 +370,7 @@ def call_chat_completion(messages, temperature=0.7, response_format=None, stream
if response_format:
completion_args["response_format"] = response_format
+ request_started_at = time.monotonic()
response = effective_client.chat.completions.create(**completion_args)
# Track usage for non-streaming calls
@@ -373,6 +399,17 @@ def call_chat_completion(messages, temperature=0.7, response_format=None, stream
except TokenBudgetExceeded:
raise
+ except APITimeoutError as e:
+ elapsed = time.monotonic() - request_started_at if 'request_started_at' in locals() else None
+ elapsed_str = f"{elapsed:.1f}s" if elapsed is not None else "unknown"
+ logger.error(
+ f"Chat LLM request timed out after {elapsed_str} (configured read_timeout={LLM_REQUEST_TIMEOUT}s, "
+ f"max_retries={LLM_MAX_RETRIES}, model={model_name}). "
+ f"If the elapsed time is much shorter than the configured timeout, an upstream proxy or the "
+ f"provider is closing the connection early. For reasoning models that take longer to think, "
+ f"increase LLM_REQUEST_TIMEOUT."
+ )
+ raise
except Exception as e:
logger.error(f"Chat LLM API call failed: {e}")
raise