Commands and Arguments in Kubernetes

Welcome to this lesson on handling commands and arguments within a Kubernetes pod. In this guide, you'll learn how to customize container behavior by overriding default settings defined in your Docker image. Previously, we built a simple Docker image called "ubuntu-sleeper" that pauses execution (sleeps) for a specified number of seconds. By default, running:

docker run ubuntu-sleeper

makes the container sleep for five seconds. However, you can change this behavior by providing a command-line argument. For example, running:

docker run --name ubuntu-sleeper ubuntu-sleeper 10

will cause the container to sleep for 10 seconds.

Overriding Default Arguments in a Pod Definition

Kubernetes allows you to replicate the Docker behavior of passing command-line arguments by using the args field in a pod definition. When you specify additional arguments in a Kubernetes pod, they are supplied as an array.

Consider this pod definition template. In the example below, the pod runs the "ubuntu-sleeper" container and overrides the default sleep duration by setting the args field to ["10"]:

apiVersion: v1
kind: Pod
metadata:
  name: ubuntu-sleeper-pod
spec:
  containers:
    - name: ubuntu-sleeper
      image: ubuntu-sleeper
      args: ["10"]

To create the pod, execute:

kubectl create -f pod-definition.yml

When the pod starts, it creates a container from the specified image. The args field in the Kubernetes definition effectively overrides the default CMD instruction defined in the Dockerfile.

Overriding the Entrypoint

In our Dockerfile, we defined an entry point and a CMD instruction as follows:

FROM ubuntu
ENTRYPOINT ["sleep"]
CMD ["5"]

Typically, when you run the container, the entry point sleep is combined with the CMD default 5. To override the entry point in Docker, you would use the --entrypoint flag, for instance:

docker run --name ubuntu-sleeper --entrypoint sleep2.0 ubuntu-sleeper 10

In Kubernetes, you achieve the same result by using the command field in the pod definition. Specifically, the command field replaces the ENTRYPOINT from the Dockerfile, and the args field continues to override the CMD instruction. Below is an example demonstrating how to set both fields:

apiVersion: v1
kind: Pod
metadata:
  name: ubuntu-sleeper-pod
spec:
  containers:
    - name: ubuntu-sleeper
      image: ubuntu-sleeper
      command: ["sleep2.0"]
      args: ["10"]

To deploy this pod, run:

kubectl create -f pod-definition.yml

Summary

There are two primary fields in a Kubernetes pod definition that correspond to your Dockerfile settings:

By correctly configuring these fields, you can modify the startup parameters of your container dynamically. This capability is particularly useful for customizing application behavior in different environments.

Review the provided exercises to practice configuring and troubleshooting commands and arguments in Kubernetes, and enhance your deployment strategies effectively.

For additional resources, refer to Kubernetes Documentation and Kubernetes Basics.