CODEWITHSUNDEEP
python-3.xgoogle-app-engineflaskgoogle-cloud-platformgunicorn
DevOpsIsNotForMe
DevOpsIsNotForMe

Reputation: 29

Deploying Flask app in Google App Engine resulting in "gunicorn: error: No application module specified."

Smacking my head against a desk trying to get this one to work. I'm trying to deploy a simple Flask app on gunicorn through Google App Engine. When visiting the app, I'm getting Error 500, with "gunicorn: error: No application module specified." in the logs.

My layout is roughly:

Main_Directory/
|-MyApp/
|   |-__init__.py <- Containing majority of code
|   |-templates/
|   |-static/
|-app.yaml
|-main.py 
|-requirements.txt

main.py simply imports the app from MyApp and instantiates a new instance called "app".

requirements.txt has all dependencies, including gunicorn.

app.yaml is exactly as follows:

runtime: python37
instance_class: F1
default_expiration: "4d 12h"
entrypoint: gunicorn -b :$PORT -w 1

env_variables:
  BUCKET_NAME: "bucket_name"

handlers:
- url: /.*
  script: auto

I tried creating another app.yaml within the MyApp folder and adding service: "default" but that didn't help. I'm pretty much out of ideas here. I'm fairly certain the issue is because the main app code is defined in MyApp, rather than directly under the Main_directory. I appreciate the easy solution would be to bring everything from the MyApp folder up a level but I'm hesitating to do this as it will mess with my Git. I thought creating main.py and directly instantiating MyApp/__init__.py would do the trick, but doesn't quite seem to.

Any and all ideas appreciated.

Upvotes: 1

Views: 1875

Answers (3)

Motti
Motti

Reputation: 114805

Another thing that can cause this error is if the gunicorn.conf.py file is missing.

Upvotes: 0

Dustin Ingram
Dustin Ingram

Reputation: 21580

If you want to use gunicorn as your HTTP server, you need to tell it where your WSGI app is located. This is likely an app variable in your main.py file, so you can use:

entrypoint: gunicorn -b :$PORT -w 1 main:app

Upvotes: 2

DevOpsIsNotForMe
DevOpsIsNotForMe

Reputation: 29

As is often the case, after reaching my limit and raising a question, I found an answer...

I removed the entrypoint from app.yaml and removed gunicorn from requirements.txt. On redeploying the app, it now works.

I don't feel like this is a fix, so much as a workaround. I'm still not sure why it didn't work before. There must be better solutions which maintain gunicorn as a requirement. I expect it was the yaml that was wrong.

Upvotes: 0

Related Questions