phoenix/litellm
1
0
Fork 0
forked from phoenix/litellm-mirror
Code Pull requests Activity

fix otel load test

Browse source
This commit is contained in:
Ishaan Jaff 2024-09-11 21:27:31 -07:00
parent b80f27dce3
commit 88706488f9
1 changed files with 19 additions and 30 deletions
Download patch file Download diff file Expand all files Collapse all files

49
tests/load_tests/test_otel_load_test.py
View file

@ -13,48 +13,37 @@ import pytest
def test_otel_logging_async(): def test_otel_logging_async():
# this tests time added to make otel logging calls, vs just acompletion calls
try: try:
os.environ["OTEL_EXPORTER"] = "otlp_http" os.environ["OTEL_EXPORTER"] = "otlp_http"
os.environ["OTEL_ENDPOINT"] = ( os.environ["OTEL_ENDPOINT"] = (
"https://exampleopenaiendpoint-production.up.railway.app/traces" "https://exampleopenaiendpoint-production.up.railway.app/traces"
) )
os.environ["OTEL_HEADERS"] = "Authorization=K0BSwd" os.environ["OTEL_HEADERS"] = "Authorization=K0BSwd"
# Make 5 calls with an empty success_callback def single_run():
litellm.success_callback = [] litellm.callbacks = []
litellm.callbacks = [] start_time_empty = asyncio.run(make_async_calls())
litellm._async_success_callback = [] print(f"Time with empty callback: {start_time_empty}")
litellm._async_failure_callback = []
litellm._async_failure_callback = []
litellm.failure_callback = []
start_time_empty_callback = asyncio.run(make_async_calls())
print("done with no callback test")
print("starting otel test") litellm.callbacks = ["otel"]
# Make 5 calls with success_callback set to "otel" start_time_otel = asyncio.run(make_async_calls())
litellm.callbacks = ["otel"] print(f"Time with otel callback: {start_time_otel}")
start_time_otel = asyncio.run(make_async_calls())
print("done with otel test")
# Compare the time for both scenarios percent_diff = (
print(f"Time taken with success_callback='otel': {start_time_otel}") abs(start_time_otel - start_time_empty) / start_time_empty * 100
print(f"Time taken with empty success_callback: {start_time_empty_callback}") )
print(f"Run performance difference: {percent_diff:.2f}%")
return percent_diff
# Calculate the percentage difference percent_diffs = [single_run() for _ in range(3)]
percentage_diff = ( avg_percent_diff = sum(percent_diffs) / len(percent_diffs)
abs(start_time_otel - start_time_empty_callback)
/ start_time_empty_callback print(f"Percentage differences: {percent_diffs}")
* 100 print(f"Average performance difference: {avg_percent_diff:.2f}%")
)
# Assert that the difference is not more than 10%
assert ( assert (
percentage_diff < 10 avg_percent_diff < 10
), f"Performance difference of {percentage_diff:.2f}% exceeds 10% threshold" ), f"Average performance difference of {avg_percent_diff:.2f}% exceeds 10% threshold"
print(f"Performance difference: {percentage_diff:.2f}%")
except litellm.Timeout as e: except litellm.Timeout as e:
pass pass