Merge pull request #1399 from BerriAI/litellm_default_use_gunicorn

LiteLLM Proxy - Use Gunicorn with Uvicorn workers
This commit is contained in:
Ishaan Jaff 2024-01-10 21:46:04 +05:30 committed by GitHub
commit 58d0366447
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
7 changed files with 62 additions and 17 deletions

View file

@ -38,6 +38,7 @@ jobs:
pip install openai
pip install prisma
pip install "httpx==0.24.1"
pip install "gunicorn==21.2.0"
pip install "anyio==3.7.1"
pip install "asyncio==3.4.3"
pip install "PyGithub==1.59.1"

View file

@ -42,7 +42,7 @@ def is_port_in_use(port):
@click.command()
@click.option("--host", default="0.0.0.0", help="Host for the server to listen on.")
@click.option("--port", default=8000, help="Port to bind the server to.")
@click.option("--num_workers", default=1, help="Number of uvicorn workers to spin up")
@click.option("--num_workers", default=1, help="Number of gunicorn workers to spin up")
@click.option("--api_base", default=None, help="API base URL.")
@click.option(
"--api_version",
@ -344,9 +344,10 @@ def run_server(
)
try:
import uvicorn
import gunicorn.app.base
except:
raise ImportError(
"Uvicorn needs to be imported. Run - `pip install uvicorn`"
"Uvicorn, gunicorn needs to be imported. Run - `pip 'litellm[proxy]'`"
)
if os.getenv("DATABASE_URL", None) is not None:
# run prisma db push, before starting server
@ -364,9 +365,37 @@ def run_server(
os.chdir(original_dir)
if port == 8000 and is_port_in_use(port):
port = random.randint(1024, 49152)
uvicorn.run(
"litellm.proxy.proxy_server:app", host=host, port=port, workers=num_workers
)
# Gunicorn Application Class
class StandaloneApplication(gunicorn.app.base.BaseApplication):
def __init__(self, app, options=None):
self.options = options or {} # gunicorn options
self.application = app # FastAPI app
super().__init__()
def load_config(self):
# note: This Loads the gunicorn config - has nothing to do with LiteLLM Proxy config
config = {
key: value
for key, value in self.options.items()
if key in self.cfg.settings and value is not None
}
for key, value in config.items():
self.cfg.set(key.lower(), value)
def load(self):
# gunicorn app function
return self.application
gunicorn_options = {
"bind": f"{host}:{port}",
"workers": num_workers, # default is 1
"worker_class": "uvicorn.workers.UvicornWorker",
"preload": True, # Add the preload flag
}
from litellm.proxy.proxy_server import app
StandaloneApplication(app=app, options=gunicorn_options).run() # Run gunicorn
if __name__ == "__main__":

View file

@ -41,13 +41,6 @@ model_list:
api_key: os.environ/OPENAI_API_KEY
model_info:
mode: embedding
- model_name: text-davinci-003
litellm_params:
model: text-davinci-003
api_key: os.environ/OPENAI_API_KEY
model_info:
mode: completion
litellm_settings:
fallbacks: [{"openai-gpt-3.5": ["azure-gpt-3.5"]}]
# cache: True

View file

@ -14,7 +14,6 @@ sys.path.insert(
) # Adds the parent directory to the system path - for litellm local dev
try:
import uvicorn
import fastapi
import backoff
import yaml

28
poetry.lock generated
View file

@ -1,4 +1,4 @@
# This file is automatically @generated by Poetry 1.5.1 and should not be changed by hand.
# This file is automatically @generated by Poetry 1.7.1 and should not be changed by hand.
[[package]]
name = "aiohttp"
@ -670,6 +670,26 @@ gitdb = ">=4.0.1,<5"
[package.extras]
test = ["black", "coverage[toml]", "ddt (>=1.1.1,!=1.4.3)", "mock", "mypy", "pre-commit", "pytest", "pytest-cov", "pytest-instafail", "pytest-subtests", "pytest-sugar"]
[[package]]
name = "gunicorn"
version = "21.2.0"
description = "WSGI HTTP Server for UNIX"
optional = true
python-versions = ">=3.5"
files = [
{file = "gunicorn-21.2.0-py3-none-any.whl", hash = "sha256:3213aa5e8c24949e792bcacfc176fef362e7aac80b76c56f6b5122bf350722f0"},
{file = "gunicorn-21.2.0.tar.gz", hash = "sha256:88ec8bff1d634f98e61b9f65bc4bf3cd918a90806c6f5c48bc5603849ec81033"},
]
[package.dependencies]
packaging = "*"
[package.extras]
eventlet = ["eventlet (>=0.24.1)"]
gevent = ["gevent (>=1.4.0)"]
setproctitle = ["setproctitle"]
tornado = ["tornado (>=0.2)"]
[[package]]
name = "h11"
version = "0.14.0"
@ -1238,8 +1258,8 @@ files = [
[package.dependencies]
numpy = [
{version = ">=1.20.3", markers = "python_version < \"3.10\""},
{version = ">=1.21.0", markers = "python_version >= \"3.10\""},
{version = ">=1.23.2", markers = "python_version >= \"3.11\""},
{version = ">=1.21.0", markers = "python_version >= \"3.10\" and python_version < \"3.11\""},
]
python-dateutil = ">=2.8.2"
pytz = ">=2020.1"
@ -2664,9 +2684,9 @@ testing = ["big-O", "jaraco.functools", "jaraco.itertools", "more-itertools", "p
[extras]
extra-proxy = ["streamlit"]
proxy = ["backoff", "fastapi", "orjson", "pyyaml", "rq", "uvicorn"]
proxy = ["backoff", "fastapi", "gunicorn", "orjson", "pyyaml", "rq", "uvicorn"]
[metadata]
lock-version = "2.0"
python-versions = ">=3.8.1,<3.9.7 || >3.9.7"
content-hash = "9f15083d98fe14237abea81eaca802e1db28cfb89bbe127498aa1fabb3c99849"
content-hash = "b49d09f51e8a57cdf883ab03cd9fecaf1ad007c3092d53347e30129e25adceab"

View file

@ -18,6 +18,7 @@ jinja2 = "^3.1.2"
aiohttp = "*"
uvicorn = {version = "^0.22.0", optional = true}
gunicorn = {version = "^21.2.0", optional = true}
fastapi = {version = "^0.104.1", optional = true}
backoff = {version = "*", optional = true}
pyyaml = {version = "^6.0", optional = true}
@ -27,6 +28,7 @@ streamlit = {version = "^1.29.0", optional = true}
[tool.poetry.extras]
proxy = [
"gunicorn",
"uvicorn",
"fastapi",
"backoff",

View file

@ -6,6 +6,7 @@ pydantic>=2.5 # openai req.
backoff==2.2.1 # server dep
pyyaml==6.0 # server dep
uvicorn==0.22.0 # server dep
gunicorn==21.2.0 # server dep
boto3==1.28.58 # aws bedrock/sagemaker calls
redis==4.6.0 # caching
prisma==0.11.0 # for db