0

This wiki was very helpful in providing a lot of detail about the process of deploying a custom datalab vm in the GCE.

In particular, the 'Releasing A Build' page documented the following location for the most current Datalab Docker container: gcr.io/cloud_datalab/datalab:latest.

Ultimately, for production, I'll conform to all of the release steps, particularly testing locally. Since I'm working in a Sandbox, however, I'd like to believe that I can streamline the process. That is, I'd like to simply clone the datalab:latest, make my changes, save it to my Git repo, and then use the Deployer App to create a new version (with an appropriate name) and set the container = deployer URL parameter to my customized image (i.e. Docker file).

My questions are:

  1. Does this seem like a correct and reasonable approach ?
  2. Is this the datalab:latest Docker file?
  3. Is the dockerfile.in what I need to clone and then make changes to ?
konqi
  • 5,137
  • 3
  • 34
  • 52
EcoWarrior
  • 571
  • 1
  • 4
  • 12

1 Answers1

1
  1. Yes.
  2. Yes.
  3. No, you should clone all project, because Dockerfile.in has references to projects. For example config/ipython.py. You can see this in the 116 line. (# Add build artifacts)
konqi
  • 5,137
  • 3
  • 34
  • 52
Nico
  • 858
  • 10
  • 20