genconfig.py allows you to snapshot a set of live GCP resources into a Deployment Manager Jinja template so that you can repeatably deploy those resources or share their configuration with others.

The tool takes a project, a file containing a list of resource selfLinks, and an optional output directory, in which three files will be generated. If no output directory is specified, it will use the current directory.

NOTE: genconfig.py requires that gcloud be locally available and properly configured to work.

The main input to genconfig is a file containing resource URLs. These URLs must be the selfLink URL from the Cloud resource itself, and can be found with one of the following methods:

• gcloud: fetch the full YAML form of the desired resource. An example command: gcloud compute instances describe my-instance --format yaml
• Equivalent REST: look up the equivalent JSON form of the desired resource in the Cloud Console. At the bottom of the resource details page, click the "Equivalent REST" link.

genconfig supports both the full path format and the shortened format:

• https://www.googleapis.com/compute/v1/projects/my-project/zones/us-central1-f/instanceGroupManagers/test
• projects/my-project/zones/us-central1-f/instanceGroupManagers/test

Store the resource URLs, one per line, in a text file for input.

For usage information, call genconfig with no parameters.

./genconfig.py

genconfig takes the project in which the resources live, the filename containing the resources, and an optional output directory. The output director defaults to the current working directory.

./genconfig.py my-project resources.txt output/

This will result in the following generated files:

• config.yaml: a simple top level config which instantiates the generated template.
• generated.jinja: the generated template containing the resource definitions.
• generated.jinja.schema: the schema for the generated template.

The generated Jinja template has parameterized the project in the configuration, but you may want to do some more customization.

Currently, genconfig does not fill in resource relationships, so be sure to add references between resources that have dependencies.

For example, an Instance Group Manager which depends on an Instance Template will be generated with a link,

instanceTemplate: https://www.googleapis.com/compute/v1/projects/{{env['project']}}/global/instanceTemplates/my-instance-template

but you may need to add a reference so it deploys in the correct order.

instanceTemplate: $(ref.my-instance-template.selfLink)

You may also want to add more parameterization to your new template, like which zone the resources are deployed to, or VM size. For more information, see the documentation on using templates and using schemas.

First deploy the HA Service example. This will create a load balanced autoscaled managed instance group to work with.

gcloud deployment-manager deployments create ha \
  --config ha-service/example.yaml

Fill in the appropriate resource URLs. See this example input.

Now run genconfig to generate the new templates.

./genconfig.py <project> example/resources.txt generated/

After generating the new template, be sure to add references between the resources to ensure proper deployment ordering. See this example output.

Before redeploying, delete the old resources and deployment.

gcloud deployment-manager deployments delete ha

Now you can redeploy using the new configurations.

gcloud deployment-manager deployments create ha --config example/config.yaml

• Only supports compute resources right now.
• Does not add references between resources.
• Does not work with Compute Instances, as they do not retain some important input parameters as part of the Instance object (namely, static vs. ephemeral IP, and InitializationParams for boot disk).