Rank 1: Thread
Hi team! Before anything, I am on the so far while we're in the planning phase of the app, and will be upgrading to pro soon. If there's a possibility that this is what's causing the issue below, please let me know.
I’m running into an issue deploying a Python function in Appwrite Functions. The function is intended to run an ML model called DeepSkin, which is used for wound segmentation and PWAT scoring for a healthcare client. This model requires the following dependencies:
numpy
mahotas
opencv-python
tensorflow>=2.8.0
gdown
git+https://github.com/Nico-Curti/Deepskin.git
The problem is not with my code, I’ve confirmed that everything runs fine without these dependencies. But once I include them (especially tensorflow, I suspect), I start encountering issues even when the function body just returns "hello world". In my build commands, I use pip install -r functions/requirements.txt to install these dependencies.
Specifically:
• The function builds fine, although it takes over 6 minutes each time I rebuild.
• But on cold start, I get:
Function timed out during cold start. Error Code: 400
• If I increase the timeout, I get:
Synchronous function execution timed out. Use asynchronous execution instead, or ensure the execution duration doesn't exceed 30 seconds. Error Code: 408
Again, none of the heavy ML logic is being executed yet, just simply running the function with these dependencies being installed seems to be the issue.
Is there anything I can do? I really don't want to go back to using FastAPI, and these serverless functions were one of the main reasons I came to you guys. Any help would be appreciated!
Thanks!