phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-26 03:04:13 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix(proxy/utils.py): tracking db failed writes

Browse source
This commit is contained in:
Krrish Dholakia 2023-12-08 10:40:02 -08:00
parent 1b35736797
commit 4e6a8d09d0
2 changed files with 99 additions and 78 deletions
Download patch file Download diff file Expand all files Collapse all files

32
litellm/proxy/utils.py
View file

@ -70,15 +70,25 @@ class PrismaClient:
"""
Update existing data
"""
hashed_token = self.hash_token(token=token)
data["token"] = hashed_token
await self.db.litellm_verificationtoken.update(
where={
"token": hashed_token
},
data={**data} # type: ignore
)
return {"token": token, "data": data}
try:
hashed_token = self.hash_token(token=token)
data["token"] = hashed_token
await self.db.litellm_verificationtoken.update(
where={
"token": hashed_token
},
data={**data} # type: ignore
)
print("\033[91m" + f"DB write succeeded" + "\033[0m")
return {"token": token, "data": data}
except Exception as e:
print()
print()
print()
print("\033[91m" + f"DB write failed: {e}" + "\033[0m")
print()
print()
print()
async def delete_data(self, tokens: List):
"""
@ -96,8 +106,7 @@ class PrismaClient:
async def disconnect(self):
await self.db.disconnect()
# ### CUSTOM FILE ###
### CUSTOM FILE ###
def get_instance_fn(value: str, config_file_path: Optional[str] = None) -> Any:
try:
print(f"value: {value}")
@ -134,7 +143,6 @@ def get_instance_fn(value: str, config_file_path: Optional[str] = None) -> Any:
raise e
### CALL HOOKS ###
class CallHooks:
"""
Allows users to modify the incoming request / output to the proxy, without having to deal with parsing Request body.

145
litellm/tests/test_proxy_server_spend.py
View file

@ -1,69 +1,82 @@
# import openai, json
# client = openai.OpenAI(
# api_key="sk-1234",
# base_url="http://0.0.0.0:8000"
# )
import openai, json, time, asyncio
client = openai.AsyncOpenAI(
api_key="sk-1234",
base_url="http://0.0.0.0:8000"
)
# super_fake_messages = [
# {
# "role": "user",
# "content": "What's the weather like in San Francisco, Tokyo, and Paris?"
# },
# {
# "content": None,
# "role": "assistant",
# "tool_calls": [
# {
# "id": "1",
# "function": {
# "arguments": "{\"location\": \"San Francisco\", \"unit\": \"celsius\"}",
# "name": "get_current_weather"
# },
# "type": "function"
# },
# {
# "id": "2",
# "function": {
# "arguments": "{\"location\": \"Tokyo\", \"unit\": \"celsius\"}",
# "name": "get_current_weather"
# },
# "type": "function"
# },
# {
# "id": "3",
# "function": {
# "arguments": "{\"location\": \"Paris\", \"unit\": \"celsius\"}",
# "name": "get_current_weather"
# },
# "type": "function"
# }
# ]
# },
# {
# "tool_call_id": "1",
# "role": "tool",
# "name": "get_current_weather",
# "content": "{\"location\": \"San Francisco\", \"temperature\": \"90\", \"unit\": \"celsius\"}"
# },
# {
# "tool_call_id": "2",
# "role": "tool",
# "name": "get_current_weather",
# "content": "{\"location\": \"Tokyo\", \"temperature\": \"30\", \"unit\": \"celsius\"}"
# },
# {
# "tool_call_id": "3",
# "role": "tool",
# "name": "get_current_weather",
# "content": "{\"location\": \"Paris\", \"temperature\": \"50\", \"unit\": \"celsius\"}"
# }
# ]
super_fake_messages = [
{
"role": "user",
"content": f"What's the weather like in San Francisco, Tokyo, and Paris? {time.time()}"
},
{
"content": None,
"role": "assistant",
"tool_calls": [
{
"id": "1",
"function": {
"arguments": "{\"location\": \"San Francisco\", \"unit\": \"celsius\"}",
"name": "get_current_weather"
},
"type": "function"
},
{
"id": "2",
"function": {
"arguments": "{\"location\": \"Tokyo\", \"unit\": \"celsius\"}",
"name": "get_current_weather"
},
"type": "function"
},
{
"id": "3",
"function": {
"arguments": "{\"location\": \"Paris\", \"unit\": \"celsius\"}",
"name": "get_current_weather"
},
"type": "function"
}
]
},
{
"tool_call_id": "1",
"role": "tool",
"name": "get_current_weather",
"content": "{\"location\": \"San Francisco\", \"temperature\": \"90\", \"unit\": \"celsius\"}"
},
{
"tool_call_id": "2",
"role": "tool",
"name": "get_current_weather",
"content": "{\"location\": \"Tokyo\", \"temperature\": \"30\", \"unit\": \"celsius\"}"
},
{
"tool_call_id": "3",
"role": "tool",
"name": "get_current_weather",
"content": "{\"location\": \"Paris\", \"temperature\": \"50\", \"unit\": \"celsius\"}"
}
]
# super_fake_response = client.chat.completions.create(
# model="gpt-3.5-turbo",
# messages=super_fake_messages,
# seed=1337,
# stream=False
# ) # get a new response from the model where it can see the function response
async def chat_completions():
super_fake_response = await client.chat.completions.create(
model="gpt-3.5-turbo",
messages=super_fake_messages,
seed=1337,
stream=False
) # get a new response from the model where it can see the function response
await asyncio.sleep(1)
return super_fake_response
# print(json.dumps(super_fake_response.model_dump(), indent=4))
async def loadtest_fn(n = 2000):
global num_task_cancelled_errors, exception_counts, chat_completions
start = time.time()
tasks = [chat_completions() for _ in range(n)]
chat_completions = await asyncio.gather(*tasks)
successful_completions = [c for c in chat_completions if c is not None]
print(n, time.time() - start, len(successful_completions))
# print(json.dumps(super_fake_response.model_dump(), indent=4))
asyncio.run(loadtest_fn())