谷歌云 Function Gen 1 部署失败

Question

This is mostly for the community, because it took me quite a while to figure out.

I have a python 3.9 environment stateless Google Cloud Function using pytorch. The initial deployment and function creation has no errors/issues. Requirements.txt looks like this:

torch==1.10.2

However, when I edit even the smallest, innocuous change and save, the deployment fails with the error:

Build failed:
Collecting torch==1.10.2; Error ID: c84b3231

How do you redeploy with pytorch?

Answer 1

I still do not know the reason why installing pytorch on redeploy fails, but this is how I got Google Cloud Function to consistently redeploy without fail. I directly input the specific version of PyTorch link: linux, cpu-based, version.

1. Go to the latest versions of PyTorch: PyTorch Latest Releases .
2. Search for the desired version (currently, the latest is 1.10.2). I chose torch-1.10.2%2Bcpu-cp39-cp39-linux_x86_64.whl .
3. Copy link and paste in requirements.txt by replacing torch==1.10.2 .

If anybody knows why Google Cloud Function Gen 1 rejects redeployment with the conventional distribution of torch, feel free to edit.

Hope this saves someone time.