0

I have a minimal Django project that I am trying to deploy onto Google App Engine Managed VMs. I'm having a bit of trouble understanding exactly what is going on and what is serving what. I have an app.yaml file that looks like this: 

runtime: custom
vm: true
entrypoint: custom
api_version: 1

manual_scaling:
  instances: 1

handlers:
- url: /static
  static_dir: static

- url: /.*
  script: DjangoVM.wsgi.application

And a dockerfile that looks like this:

FROM gcr.io/google_appengine/python
RUN apt-get update
RUN apt-get -y install python-numpy python-scipy
ADD requirements.txt /app/requirements.txt
RUN pip install -r /app/requirements.txt
ADD . /app
CMD gunicorn DjangoVM.wsgi

My requirements.txt looks like this:

Django==1.8.5
gunicorn==19.3.0
scikit-learn==0.16.1

The site deploys okay, but my static files aren't showing up. So on a sandboxed app engine project the handlers in the app.yaml take care of serving my static content. Do I have to configure an nginx server or similar server in my dockerfile to serve static content?

To help me understand this, if I didn't have gunicorn serving my application, does App Engine not serve it by default using the handlers? i.e. What would happen if I didn't have CMD gunicorn DjangoVM.wsgi? Does a custom runtime mean that app engine will not serve my application by default and I have to specify a particular WSGI server? The documentation is not very thorough for Managed VMs yet. Sorry if the questions seem a bit confusing, but it reflects my current confusion on the subject as well. Thanks for any help!

TowMater7
  • 111
  • 2
  • 12

1 Answers1

0

It looks like you've confused two things here.

There are two ways of deploying a managed vm: using a standard runtime, in this case Python, and using a custom runtime.

The standard runtime is basically an App Engine app; you define your routes in app.yaml as usual. The runtime value should be "python27". However, you don't need a Dockerfile, and so it wouldn't include nginx or gunicorn; the routes are served by the App Engine runtime.

A custom runtime is defined by giving a runtime value of "custom". This is a completely custom app, defined by a Dockerfile. You don't declare routes in app.yaml, and they need to be served by properly-configured nginx/gunicorn within your container.

Daniel Roseman
  • 588,541
  • 66
  • 880
  • 895
  • Thanks for clearing that up for me. A follow-up question though. If I use the standard runtime, is there a way to take advantage of the Managed VM environment by utilizing apt-get? I cloned and ran this project: https://github.com/GoogleCloudPlatform/appengine-scipy-zombie-apocalypse-python (and also used the same method in my project) but to no avail. The 'vm_settings' in app.yaml doesn't seem to be doing anything as matplotlib was not installed on the VM. I remember seeing somewhere that it was deprecated, but not sure the replacement. – TowMater7 Nov 23 '15 at 23:26
  • Custom runtime using Google base image FROM gcr.io/google_appengine/python-compat should be equivalent to using python27 runtime with managed vm (no need of nginx or gunicorn), and I guess most people use this, because this is the only way to install extra packages through pip. – dyeray Feb 11 '16 at 11:26