Often times you need to inject a configuration file into a container in Kubernetes. This is really easy to do using the ConfigMap resource. But once in a while you might need to inject an executable script into a container.
There are any number of reasons why you might need to do so. In my case, I needed to alter the default behaviour of a Docker image. Normally you might create a new git repo with a Dockerfile and the script, and then build a whole new Docker image just to include the script. Instead you can inject an executable script into the container at runtime using a ConfigMap and the
defaultMode of a volume.
The versions used in this post at the time of writing are:
Create a Kubernetes cluster with Minikube, clone the gist with example code, and run it. Note how the source of the ConfigMap is the wrapper.sh file.
The Executable Script
In this case, the script is just a wrapper around the regular entrypoint for the ghost image that allows you to do some special initialization beforehand.
A volume is created from the ConfigMap with
defaultMode: 0744, that’s what makes it executable. It’s then mounted to a /scripts dir but it could be mounted anywhere. The
command: ["/scripts/wrapper.sh"] overrides the Docker image’s entrypoint and runs wrapper.sh instead.
Before I found out about
defaultMode, my first crack at solving this problem was to use an Init Container like this.
It gets the job done but
defaultMode is a much more elegant and succinct way to do it.
You can inject any kind of text based file into a container in Kubernetes. Making it an executable script is just one special case of that. I like this approach a lot because you don’t have to create and maintain yet another Docker image just to inject one particular file.
P.S. I discovered that this is also extremely useful for extending the official Postgres image and creating scripts in the /docker-entrypoint-initdb.d/ dir of the Postgres container.