Deploying models to Google App Engine Flexible

deployment
1

I’m trying to deploy a Torch/Transformers model on Google App Engine Flexible. During deploy, I get this error:

Updating service [default] (this may take several minutes)...failed.                                                                      
ERROR: (gcloud.app.deploy) Error Response: [4] Your deployment has failed to become healthy in the allotted time and therefore was rolled back. If you believe this was an error, try adjusting the 'app_start_timeout_sec' setting in the 'readiness_check' section.

I’ve done the recommended change and also cranked up the instance memory to 10GB but I still get the same error. I’m copying my app.yaml below.

Has anyone successfully deployed a model like this to GAE flex?

I’m able to deploy to AWS but need it on GAE flex…

=====

runtime: python
env: flex

entrypoint: gunicorn -b :$PORT main:app

runtime_config:
  python_version: 3

manual_scaling:
  instances: 2

resources:
  cpu: 1
  memory_gb: 10
  disk_size_gb: 10

network:
  name: default

readiness_check:
  app_start_timeout_sec: 1800
2

Increasing the disk size to 20gb fixed this for me. There was actually an error message in the deploy logs but it is easy to miss because of the sheer volume of log messages.

1 Like
3

Hey Jeff I am glad to see you this. I am having same trouble installing pytorch into my app engine application. Can you please share your cloudbuild as well. Whenever I use runtime: Python the build error out.