What the best way to close server in case of an Exception

šŸ”’ Gradio
1

Iā€™m trying to deploy an application using Gradio, but Iā€™m using kubernetes. My main issue is that, when something does not work with the inference function, the app is not shut down as it should. Because of that, the pod continues to run and another new pod is not created.

What is the best way to handle this issue?

import gradio as gr
import pandas as pd

from main_module.main import run_inference


def read_file(file_object):
    """
    read input file
    """
    df = pd.read_excel(file_object.name, engine="openpyxl")

    output: pd.DataFrame = run_inference(df)

    file_output_path = "result.xlsx"
    with open(file_output_path, "wb") as file:
        output.to_excel(file, index=False, encoding="utf-8")
        return file_output_path


def create_gradio():
    """
    create gradio app
    """
    iface = gr.Interface(fn=read_file, inputs="file", outputs="file")
    try:
        iface.launch(server_name="0.0.0.0", server_port=8080)
    except KeyboardInterrupt:
        iface.close()
    except Exception as e:
        print(e)
        iface.close()


if __name__ == "__main__":
    create_gradio()
2

Hmm the way that you have it should work. The Interface.close() method is designed to close the server and release the port.

If this does not work, you might want to try something more heavy-handed to free the port: e.g. Is it possible in python to kill process that is listening on specific port, for example 8080? - Stack Overflow