phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-27 11:43:54 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

(fix) test_normal_router_tpm_limit

Browse source
This commit is contained in:
ishaan-jaff 2024-02-06 12:44:30 -08:00 • committed by Krrish Dholakia
parent bdb9fc8f20
commit adb67da026
1 changed files with 6 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

6
litellm/tests/test_parallel_request_limiter.py
View file

@ -306,6 +306,10 @@ async def test_normal_router_call():
@pytest.mark.asyncio @pytest.mark.asyncio
async def test_normal_router_tpm_limit(): async def test_normal_router_tpm_limit():
from litellm._logging import verbose_proxy_logger
import logging
verbose_proxy_logger.setLevel(level=logging.DEBUG)
model_list = [ model_list = [
{ {
"model_name": "azure-model", "model_name": "azure-model",
@ -353,6 +357,7 @@ async def test_normal_router_tpm_limit():
current_minute = datetime.now().strftime("%M") current_minute = datetime.now().strftime("%M")
precise_minute = f"{current_date}-{current_hour}-{current_minute}" precise_minute = f"{current_date}-{current_hour}-{current_minute}"
request_count_api_key = f"{_api_key}::{precise_minute}::request_count" request_count_api_key = f"{_api_key}::{precise_minute}::request_count"
print("Test: Checking current_requests for precise_minute=", precise_minute)
assert ( assert (
parallel_request_handler.user_api_key_cache.get_cache( parallel_request_handler.user_api_key_cache.get_cache(
@ -366,6 +371,7 @@ async def test_normal_router_tpm_limit():
model="azure-model", model="azure-model",
messages=[{"role": "user", "content": "Write me a paragraph on the moon"}], messages=[{"role": "user", "content": "Write me a paragraph on the moon"}],
metadata={"user_api_key": _api_key}, metadata={"user_api_key": _api_key},
mock_response="hello",
) )
await asyncio.sleep(1) # success is done in a separate thread await asyncio.sleep(1) # success is done in a separate thread
print(f"response: {response}") print(f"response: {response}")