anaconda devcontainer docs example does not work

[wkhatch]

If using the default Dockerfile, and devcontainer.json including three simple enhancements, which are also mentioned and shown in the readme:

    // Use 'forwardPorts' to make a list of ports inside the container available locally.
"forwardPorts": [8888],


// Use 'postStartCommand' to run commands after the container is created.
"postStartCommand": "nohup bash -c 'jupyter notebook --ip=0.0.0.0 --port=8888 --allow-root &'",


// Set `remoteUser` to `root` to connect as root instead. More info: https://aka.ms/vscode-remote/containers/non-root.
"remoteUser": "vscode"

This will never complete when opening a local workspace in devcontainer.

I'm not sure what the fix is, but, I think the example should be made to work correctly, as this is a very common configuration.

[Kaniska244]

Hello @wkhatch ,

Thank you for raising this issue. Would you kindly try with "appPort": [8888], option instead of "forwardPorts": [8888], in the devcontainer.json & let me know if you are still unable to open the jupyter notebook locally in the browser.

With Regards,
Kaniska

[Kaniska244]

Hello @wkhatch ,

On a second look the "forwardPorts": [8888] works fine for me when using with VS code devcontainers extension in VSCode & I am able to locally open the jupyter notebook in browser based on the URL provided which is running on port 8888 inside the container.
The above I had mentioned about "appPort": [8888] is applicable when using devcontainers cli which currently doesn’t support forwardPorts tag, there is an open issue for this.

Would you kindly give me some more details on the issue you are facing in this?

With Regards,
Kaniska

[wkhatch]

@Kaniska244

Apologies for the delay. I tried using appPort, and still have same issue, where opening in the container fails. I have been using these files: https://github.com/devcontainers/images/tree/main/src/anaconda by copying all of them into my .devcontainer directory. The Dockerfile won't build due to some issues with apply_security_patches.sh and add-notice.sh. I didn't catch those at first, because when it fails, and you "Reopen Locally" I would lose the debug output. After figuring that out, I was able to get a little further, but not much, so, I'm still failing to "Reopen in Container" because the Dockerfile won't build. Thanks for help.

Oh, also, the reason I'm trying these files, as opposed to just using the standard files that get generated, is that I was not able to load an existing environment, nor was I able to do it manually using conda, once the container came up. I would get the dreaded "try conda init... first... " error. Also, the stock files from generator do not launch jupyter lab, although if I could get through the conda init issue, I'd be able to easily launch it from the shell. Hope that provides additional context. Thanks again.

[Kaniska244]

Hi @wkhatch ,

Thank you for the details. I tried these steps to activate a new conda environment from the devcontainer with anaconda and jupyter notebook. Would you kindly have a look & let me know if I am missing something on the actual issue.

I used the following devcontainer.json for this project.

{
    "image": "mcr.microsoft.com/devcontainers/anaconda:dev-3",
    "forwardPorts": [8888],
    "postStartCommand": "nohup bash -c 'jupyter notebook --ip=0.0.0.0 --port=8888 --allow-root &'",
    "customizations": {
        "vscode": {
            "extensions": [
                "mads-hartmann.bash-ide-vscode",
                "dbaeumer.vscode-eslint",
                "EditorConfig.EditorConfig",
                "ms-vscode-remote.remote-containers"
            ]
        }
    }
}

Then once the devcontainer came up, inside that I used conda create --name anothertestenv -c conda-forge --yes python=3.13 numpy pandas matplotlib scikit-learn mistune transformers cryptography=44.0.1 jupyter-lsp scrapy zipp tornado ipykernel command to create a new environment called anothertestenv.

Followed by that while trying to activate the conda environment, I encountered CondaError: Run 'conda init' before 'conda activate' error, so ran a conda init & exited the terminal.

vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ conda activate anothertestenv                                                        
                                                                                                                                                       
CondaError: Run 'conda init' before 'conda activate'                                                                                                   
                                                                                                                                                       
vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ conda list env
# packages in environment at /opt/conda:                                                                                                               
#
# Name                    Version                   Build  Channel
pylint-venv               3.0.3           py312hd43f75c_0  
python-dotenv             0.21.0          py312hd43f75c_0  
vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ conda init
no change     /opt/conda/condabin/conda
no change     /opt/conda/bin/conda
no change     /opt/conda/bin/conda-env
no change     /opt/conda/bin/activate
no change     /opt/conda/bin/deactivate
no change     /opt/conda/etc/profile.d/conda.sh
no change     /opt/conda/etc/fish/conf.d/conda.fish
no change     /opt/conda/shell/condabin/Conda.psm1
no change     /opt/conda/shell/condabin/conda-hook.ps1
no change     /opt/conda/lib/python3.12/site-packages/xontrib/conda.xsh
no change     /opt/conda/etc/profile.d/conda.csh
modified      /home/vscode/.bashrc

==> For changes to take effect, close and re-open your current shell. <==

vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ exit

Reconnected to the terminal again & was able to activate the new environment & made further changes to make this new environment visible as a kernel in the Jupyter notebook running in the devcontainer.

(base) vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ conda list env
# packages in environment at /opt/conda:
#
# Name                    Version                   Build  Channel
pylint-venv               3.0.3           py312hd43f75c_0  
python-dotenv             0.21.0          py312hd43f75c_0  
(base) vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ conda activate anothertestenv
(anothertestenv) vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ conda list env
# packages in environment at /opt/conda/envs/anothertestenv:
#
# Name                    Version                   Build  Channel
(anothertestenv) vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ jupyter server list
Currently running servers:
http://668ae6e4049e:8888/?token=b00e91545c2d29e7ce0399dc6b9fd74c87973ca582af2346 :: /workspaces/images-8
(anothertestenv) vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ 
(anothertestenv) vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ conda list ipykernel
# packages in environment at /opt/conda/envs/anothertestenv:
#
# Name                    Version                   Build  Channel
ipykernel                 6.29.5             pyh3099207_0    conda-forge
(anothertestenv) vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ python -m ipykernel install --user --name=anothertestenv
Installed kernelspec anothertestenv in /home/vscode/.local/share/jupyter/kernels/anothertestenv
(anothertestenv) vscode ➜ /workspaces/images-8 (anaconda-jupyter-notebook-issue) $ 

Now Its available in the jupyter notebook running the local browser.

[wkhatch]

Thanks; I'm not sure what you're using for a Dockerfile, and my original point is that the image and default files generated when you select Python Anaconda 3 as your devcontainer image, simply do not work. That's still the case. I've decided to abandon using anaconda, as I don't see anything but downside with it. Why anybody would want to use this seems completely irrational to me, given my (admittedly) anecdotal experiences thus far. But, researching this issue highlights that this has been a thing for at least six or seven years now, and, still a thing today. No, thank you... Either way, a user who selects the defaults for Python with Anaconda 3 as their dev container, should end up with a working devcontainer, and that is simply just not happening... Thanks for all your input and help. I really do appreciate it, even though I'm obviously sour on anaconda at the moment lol

[Kaniska244]

Hello @wkhatch ,

I had actually tried out that example without any Dockerfile, directly referring to anaconda devcontainer image (which is built from the anaconda image) from microsoft container registry in the devcontainer.json. Let me check with the anaconda image Dockerfile directly & get back to you.

With Regards,
Kaniska