Prevent failed entrypoints from spoiling the launch

[banesullivan-kobold]

I am in a JupyterLab environment where many servers are registered under the jupyter-server-proxy entrypoints and occassionaly one of them might fail. Trouble is that if one fails, they all fail and no servers are able to be registered and launched (or maybe just subsequent entrypoints fail to register?).

I'd like for jupyter-server-proxy to gracefully handle failures when loading entrypoints so as to not let one failed entrypoint prevent others from loading.

edit: See demonstration of issue in later comment

Looking forward to any feedback on these changes and if you all think this would do it

[welcome]

Thanks for submitting your first pull request! You are awesome! 🤗

If you haven't done so already, check out Jupyter's Code of Conduct.

You can meet the other Jovyans by joining our Discourse forum. There is also a intro thread there where you can stop by and say Hi! 👋

Welcome to the Jupyter community! 🎉

[yuvipanda]

Hi @banesullivan-kobold!

Am curious - do you have a stacktrace that can help me better understand what kinda failures happen here?

[banesullivan-kobold]

Here is an example.

Take the following app.py module that registers two

def setup_app():
   return {
        # Run a simple file server
        "command": ["python", "-m", "http.server", "{port}"],
        "port": 5555,
        "launcher_entry": {
            "enabled": True,
            "title": "My App",
        },
   }


def setup_broken():
   raise NotImplementedError

Then the following setup.py to install and register those servers with jupyter-server-proxy:

import setuptools

setuptools.setup(
    name="app",
    py_modules=["app"],
    entry_points={
        "jupyter_serverproxy_servers": [
            "myapp = app:setup_app",
            "broken = app:setup_broken",
        ]
    },
    install_requires=["jupyter-server-proxy"],
)

To reproduce the issue:

1. pip install -e . this app.py with its setup.py
2. Launch jupyterlab: jupyter lab
3. Notice the following error in the logs
4. Notice that on the launcher page in JupyterLab, we are missing the entry to the working app.

When launching jupyter lab via jupyter lab on the command line, we see the following error in the logs:

[W 2024-06-11 21:58:08.028 ServerApp] jupyter_server_proxy | extension failed loading with message: NotImplementedError()
    Traceback (most recent call last):
      File "/opt/conda/lib/python3.11/site-packages/jupyter_server/extension/manager.py", line 356, in load_extension
        extension.load_all_points(self.serverapp)
      File "/opt/conda/lib/python3.11/site-packages/jupyter_server/extension/manager.py", line 228, in load_all_points
        return [self.load_point(point_name, serverapp) for point_name in self.extension_points]
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
      File "/opt/conda/lib/python3.11/site-packages/jupyter_server/extension/manager.py", line 228, in <listcomp>
        return [self.load_point(point_name, serverapp) for point_name in self.extension_points]
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
      File "/opt/conda/lib/python3.11/site-packages/jupyter_server/extension/manager.py", line 219, in load_point
        return point.load(serverapp)
               ^^^^^^^^^^^^^^^^^^^^^
      File "/opt/conda/lib/python3.11/site-packages/jupyter_server/extension/manager.py", line 147, in load
        return loader(serverapp)
               ^^^^^^^^^^^^^^^^^
      File "/opt/conda/lib/python3.11/site-packages/jupyter_server_proxy/__init__.py", line 48, in _load_jupyter_server_extension
        server_processes += get_entrypoint_server_processes(serverproxy_config)
                            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
      File "/opt/conda/lib/python3.11/site-packages/jupyter_server_proxy/config.py", line 100, in get_entrypoint_server_processes
        server_process_config = entry_point.load()()
                                ^^^^^^^^^^^^^^^^^^^^
      File "/app/app.py", line 14, in setup_broken
        raise NotImplementedError
    NotImplementedError

Result	Expected
Jupyter-server-proxy entirely failed to launch and is not useable	I expect jupyter-server-proxy to continue proxying ports and I expect named servers with working configurations to continue to be registered

It is clear that having one single launch config break or be incorrectly specified will entirely break all of jupyter-server-proxy

[banesullivan-kobold]

@yuvipanda, I'm curious what I can do to help get this over the finish line? It feels like a pretty critical bug that having any package in the environment incorrectly configuring a named server would prevent all of jupyter-server-proxy from working.

Also, there is a CI failure here now that seems unrelated?

[yuvipanda]

Sorry for the delayed response, @banesullivan-kobold. I agree this is an important fix to get out. We've fallen behind on review a little, and I'm trying to get back on track.

Thank you for your contribution, and I hope future contributions will get reviewed faster!

the CI failure was unrelated, I restarted CI and it went away

[banesullivan-kobold]

Fantastic, thank you for your time reviewing and merging this!!

[consideRatio]

Updated title to reflect suggestion from Simon reviewing the changelog in #488