Skip to content

Latest commit

 

History

History
27 lines (16 loc) · 2.46 KB

README.md

File metadata and controls

27 lines (16 loc) · 2.46 KB
Raw

Docker image for dynamic template processing

This image builds on top of a statically compiled gomplate which uses scratch as base image (so it's quite small). It adds support for bulk operation and is a nicely fit for an init container to pre-process configuration files in a Kubernetes pod.

It is used in our Kubernetes Patterns book in examples for the Configuration Template pattern.

You use this image typically as a base image like in

FROM k8spatterns/gomplate
COPY myconfig/ in/

So, you typically copy only your gomplate templates into the /in directory so that they are directly available in the image. The parameters filled into the templates are filled in during runtime by mounting a volume to /params from gomplate datasource files are picked up. These datasources are referenced from within the template directory. The processed files are then written to /out. This directory should be mounted as volume, too, so that these result files are accessible from the outside.

The entry point of this image is "/gomplate" and can be configured in various ways by providing custom arguments as CMD in a sub-image. All options available can be taken from gomplate, but for our use case the most important are:

  • The --input-dir directory holds the templates. These are typically baked into the image like in this example.
  • The --datasource is the directory containing the gomplate datasources. Each file in this directory is taken as an individual datasource which is used with the file's basename as key. E.g. a config.yml can be referenced from the template as {{ (datasource "config").someKeyFromYamlFile }}. This directory is typically mounted from the outside (e.g. from a Kubernetes ConfigMap backed volume).
  • The --output-dir directory where to store the processed files. The generated files will have the same name as the input files. This directory is typically also pointing to a directory within a volume so that the processed files can be used from the application to configure.

For a full list of options see also docker run -it k8spatterns/gomplate -h.

Recreate image

You need a Docker version with support multi stage builds which are used here to compile gomplate fresh. You can use build.sh for recreating the image.