Skip to content

Latest commit

 

History

History
93 lines (79 loc) · 3.55 KB

2_4_1_Web_Server.md

File metadata and controls

93 lines (79 loc) · 3.55 KB
Raw

Web Server

AgentUniverse provides a web server based on Flask. You can call the developed Agent services through the preset APIs in the web server.

Web Server Launch

You need to start the web server after launching the AgentUniverse application. Here is an example of how to launch it:

from agentuniverse.agent_serve.web.web_booster import start_web_server
from agentuniverse.base.agentuniverse import AgentUniverse

AgentUniverse().start()
start_web_server()

Flask server default port is 8888，you can change it by edit bind param:

from agentuniverse.agent_serve.web.web_booster import start_web_server
from agentuniverse.base.agentuniverse import AgentUniverse

AgentUniverse().start()
start_web_server(bind="127.0.0.1:8002")

If you need to configure more details or pursue production stability, you can use Gunicorn as the HTTP server.

Gunicorn Configuration

You can set Gunicorn as your HTTP server by configuring it in the configuration file as follows:

[GUNICORN]
activate = 'true'

The configuration of the Gunicorn server consists of four levels in order of increasing priority: native Gunicorn default configuration, AgentUniverse default configuration, configuration file settings, and startup configuration. For Gunicorn's native default configuration, please refer to the official documentation for the corresponding version of Gunicorn.

AgentUniverse Default Configuration

DEFAULT_GUNICORN_CONFIG = {
    'bind': '127.0.0.1:8888',
    'workers': 5,
    'backlog': 2048,
    'worker_class': 'gthread',
    'threads': 4,
    'timeout': 60,
    'keepalive': 10
}

AgentUniverse has pre-configured some parameters based on the framework's working characteristics, which you can modify through the configuration file below or parameters when launching the web server.

Configuration File Path

The configuration path for the web server is specified in AgentUniverse's main configuration file config.toml.

[GUNICORN]
# Gunicorn config file path, an absolute path or a relative path based on the dir where the current config file is located.
gunicorn_config_path = './gunicorn_config.toml'

If this parameter is in relative path format, the corresponding parent directory is the directory where the main configuration file config.toml is located.

Configuration File

[GUNICORN_CONFIG]
bind = '127.0.0.1:8888'
backlog = 2048
worker_class = 'gthread'
threads = 4
workers = 5
timeout = 60
keepalive = 10

The configuration file is written in TOML format, and the specific configuration items are not limited to the content in the example above. You can refer to the official Gunicorn documentation for more detailed configurations.

Startup Configuration

You can also configure the web server at startup as follows:

from agentuniverse.agent_serve.web.web_booster import start_web_server
from agentuniverse.base.agentuniverse import AgentUniverse

AgentUniverse().start()
start_web_server(bind="127.0.0.1:8002", threads=4)

If you have certain functions you want to execute after gunicorn fork child processes, you can add them in a way similar to the following:

from agentuniverse.agent_serve.web.post_fork_queue import add_post_fork
from agentuniverse.agent_serve.web.web_booster import start_web_server
from agentuniverse.base.agentuniverse import AgentUniverse

def print_hello(name: str):
    print(f"hello {name}")

add_post_fork(print_hello, "name")
AgentUniverse().start()
start_web_server(bind="127.0.0.1:8002")

You should observe the same amount output "hello name" as gunicorn workers。