Use a python project in the storage for serverless workers | Runpod | Page 1

cursive torrent Mar 2, 2026, 2:17 AM

#

Hello, I have prepared a Python project in a pod and packed it together with all dependencies into a PIXI environment. Everything is stored together in a runpod storage, making it very quickly available and ready to start in new pods. I have now also made the project serverless ready with rp_handler.py, but I can't get it to work with the serverless command (/runpod-volume/.pixi/envs/default/bin/python3 -u /runpod-volume/myproject/rp_handler.py) that refers to the runpod storage, since I also need to use Python from my PIXI environment in storage. Alternatively, I could also build the Pixi Environment in a Docker container, but I wanted to keep the solution lean, and it is also very convenient to keep the project in storage and be able to edit and test it in a pod.

topaz crestBOT Mar 2, 2026, 2:17 AM

#

rich sleetBOT Mar 2, 2026, 2:17 AM

#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

pale osprey Mar 2, 2026, 2:48 AM

#

cursive torrent Hello, I have prepared a Python project in a pod and packed it together with all...

Any errors that you get?
I'm not familiar with pixi but I'll try to help when you give any error or more context.

cursive torrent Mar 2, 2026, 3:11 AM

#

the serverless worker was initializing for 2 hours yesterday, thats why i think the initial command is not working. No error messages in the webui. I will try to get more informations with the runpod cli today

pale osprey Mar 2, 2026, 3:30 AM

#

Check the worker logs, maybe there's some clues there

#

And are you using the model slot? (custom hf models) in the endpoint setting

#

I think runpod cli wont help out much, what will you do with it?

cursive torrent Mar 2, 2026, 7:43 AM

#

Unfortunately, the worker logs are empty and the worker is stuck on initialising, and the serverless project status is also waiting for build. - i have selected a gpu with high supply and 1 worker

#

im not using the hf models slot because i have cached a lot of models in the storage, as far as i know the model slot allows only one hf model.

whoops i thought i can get additional informations about the initializing status of the worker through runpodcli, because all the logs are empty :/

cursive torrent Mar 2, 2026, 8:16 AM

#

ok i saw a build eror and fixed it but it takes a long time until the build is repeated

#Use a python project in the storage for serverless workers