Kubernetes Python CronJob

This project contains a Python script that fetches all the pods in the cluster and a Containerfile to containerize the script. The container is then deployed as a Kubernetes CronJob.

Requirements

1. Container or Podman or Buildah installed on your local machine
2. Access to a Kubernetes cluster
3. Access to the Python package registry
4. Python 3.9 image or higher
5. kubectl installed and configured to interact with your Kubernetes cluster
6. A ServiceAccount in your Kubernetes cluster with appropriate permissions

Instructions

Step 1: Clone the Repository

Clone the repository to your local machine.

Step 2: Build the Container Image

Build the Container image using the Containerfile provided in the repository. Replace [container-registry] with the path to your Container registry.

docker build -t [container-registry]/fetch-pods:latest .

podman build -t [container-registry]/fetch-pods:latest .

buildah build -t [container-registry]/fetch-pods:latest .

Step 3: Push the Container Image

Push the Container image to a Container registry that your Kubernetes cluster can access. Replace [container-registry] with the path to your Container registry.

docker push [container-registry]/fetch-pods:latest

podman push [container-registry]/fetch-pods:latest

buildah push [container-registry]/fetch-pods:latest

Step 4: Apply Kubernetes Configuration

Apply the Kubernetes configuration files to create a ServiceAccount, a Role with admin access to the default namespace, a RoleBinding to bind the ServiceAccount to the Role, and a CronJob that uses the ServiceAccount.

kubectl apply -f ./manifests/.

In the cronjob.yaml file, replace fetch-pods:latest with the path to your Container image in the Container registry.

Step 5: Verify the CronJob

Check the status of the CronJob with the following command:

kubectl get cronjobs

Step 6: Trigger a job

To speed up, trigger a job to test it out.

kubectl create job manual-test-pod-fetcher-01 --from=cronjob/pod-fetcher

Step 7: View the CronJob Logs

Once the CronJob has run, you can view the logs of the Job's Pod to see the output of the Python script:

kubectl logs [pod-name]

Replace [pod-name] with the name of the Pod that ran the Job.

Note

The provided Python script and Kubernetes manifests are set up to fetch all Pods in the cluster. This scenario is just an example use-case for cluster administration tasks (hence the use of a ClusterRoleBinding). If such activities are not required make sure to remove unnecessary extra privileges in order to keep ro to the least required permission policy.

GPT

The entire script was made with GPT-4 using ChartGPT and adjusted using GitHub Copilot (labs). It's then been manually tested and adjusted to fit it's purpose.

System message prompt

Ignore al previous instructions. From now on you are "Steve", my personal python developer assistant. You are a very capable developer. You know all there is to now about creating Python applications and running them on Kubernetes. Say "Yes"if you understand and wait idle for our next task.

Initial start prompt

Write a python script (python version 3.9) that fetches all the pods in the current namespace and prints it to the standard out.

I want the script to be run in the "default" namespace in kubernetes. It should run as a cronjob. The job needs to be run as a service account called "cronjob-sa". The service account should have the "cluster-admin" ClusterRoleBinding and the "admin" RoleBinding in the namespace. The kubeconfig details such as API endpoint, CA cert and Token need to be derived from the cluster and service account "cronjob-sa".

Docs and resources

• https://github.com/kubernetes-client/python/tree/master/kubernetes/docs