phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-25 18:54:30 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

v0 basic structure

Browse source
This commit is contained in:
ishaan-jaff 2024-01-25 09:58:43 -08:00 • committed by Krrish Dholakia
parent 72275ad8cb
commit 565531fe9e
2 changed files with 36 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

26
litellm/proxy/proxy_server.py
View file

@ -407,6 +407,14 @@ async def user_api_key_auth(
user_max_budget is not None
and user_current_spend is not None
):
asyncio.create_task(
proxy_logging_obj.budget_alerts(
user_max_budget=user_max_budget,
user_current_spend=user_current_spend,
type="user_and_proxy_budget",
user_info=_user,
)
)
if user_current_spend > user_max_budget:
raise Exception(
f"ExceededBudget: User {valid_token.user_id} has exceeded their budget. Current spend: {user_current_spend}; Max Budget: {user_max_budget}"
@ -422,6 +430,15 @@ async def user_api_key_auth(
user_max_budget is not None
and user_current_spend is not None
):
asyncio.create_task(
proxy_logging_obj.budget_alerts(
user_max_budget=user_max_budget,
user_current_spend=user_current_spend,
type="user_budget",
user_info=user_id_information,
)
)
if user_current_spend > user_max_budget:
raise Exception(
f"ExceededBudget: User {valid_token.user_id} has exceeded their budget. Current spend: {user_current_spend}; Max Budget: {user_max_budget}"
@ -448,6 +465,15 @@ async def user_api_key_auth(
# Check 4. Token Spend is under budget
if valid_token.spend is not None and valid_token.max_budget is not None:
asyncio.create_task(
proxy_logging_obj.budget_alerts(
user_max_budget=valid_token.max_budget,
user_current_spend=valid_token.spend,
type="token_budget",
user_info=valid_token,
)
)
if valid_token.spend > valid_token.max_budget:
raise Exception(
f"ExceededTokenBudget: Current spend for token: {valid_token.spend}; Max Budget for Token: {valid_token.max_budget}"

10
litellm/proxy/utils.py
View file

@ -181,6 +181,14 @@ class ProxyLogging:
level="Low",
)
async def budget_alerts(
self,
type: Literal["token_budget", "user_budget", "user_and_proxy_budget"],
user_max_budget: float,
user_current_spend: float,
):
pass
async def alerting_handler(
self, message: str, level: Literal["Low", "Medium", "High"]
):
@ -191,6 +199,8 @@ class ProxyLogging:
- Requests are hanging
- Calls are failing
- DB Read/Writes are failing
- Proxy Close to max budget
- Key Close to max budget
Parameters:
level: str - Low|Medium|High - if calls might fail (Medium) or are failing (High); Currently, no alerts would be 'Low'.