Is Conda-Forge not Supported ?!

[Shaheer]

PLEASE HELP ME UNDERSTAND.

Here is the link to my github repo if you would like a more detailed analysis: GitHub - ShaheerSyed/mdsimulation-streamlit

I just want to know if we can or cannot deploy applications on the community cloud if the application uses conda environment packages.

If you look at my repo, I have included an environment.yml file. I don’t think there is any issue with the repo or code itself.

The error I got when trying to deploy is below:

Collecting package metadata (repodata.json): ...working... [2023-12-03 01:23:24.253950] bash: line 3:    11 Killed                  /home/adminuser/.conda/bin/conda env update -n base --file environment.yml

[01:23:24] ❗️ installer returned a non-zero exit code

[01:23:24] ❗️ Error during processing dependencies! Please fix the error and push an update, or try restarting the app.

[01:24:57] ❗️ Streamlit server consistently failed status checks

[01:24:57] ❗️ Please fix the errors, push an update to the git repo, or reboot the app.

[dataprofessor]

Hi @Shaheer

You can manage library dependencies for apps deployed on the Community Cloud via various methods such as pip, conda, poety, pipenv, etc.

In your case, you can install conda dependencies by defining them in an environment.yml file.

Please see the Docs for more info:

App dependencies - Streamlit Docs

Hope this helps!

[edsaac]

@Shaheer, yes but actually no. conda is supported but the moment you use conda-forge as one of your channels, Community Cloud will fail because it quickly spends all the available resources in the server. Here are a couple of related threads:

Error when deploying using a conda environment.yml

You are probably aware that solving an environment using conda-forge can require an awful lot of memory. So, if you have such memory requirements, then a free hosting service with only 1 GB of memory available is useless to you. But you can still deploy your Streamlit application in another, more powerful hosting.

Error during processing dependencies using environment.yml file

I just realised I didn´t explain why this happens. According the ticket, I posted earlier, conda is caching the channels. Since the conda-forge is pretty big, the cache increases the memory consumption during installation too much and the install process is killed by the kernel.

[evan54]

can one just pay for more resources to see if this resolves conda-forge issues?

[Caroline]

Unfortunately, we don’t offer paid plans on Community Cloud. If you have a deployed app that you suspect is hitting the memory limit, please share the URL and I can confirm for you.

If your app is used by an educational or non-profit organization, we can help out with more resources – in that case, please submit this form.

[George_Merticariu]

Hi! With the latest release Community Cloud, we changed the conda resolver to libmamba, which has different resource requirements than the default conda one. Therefore, conda-forge can be used as an alternative channel.