Kubernetes Image Usage Report Script

README.md

Kubernetes Image Usage Report Script

This script is designed to generate a comprehensive report of all images used in a Kubernetes cluster and to count how many times each image is used.

Author

Ariel Lothlorien

Date Created

2021-08-26

Last Updated

2021-11-02

Purpose

To retrieve and analyze a list of all images deployed in a Kubernetes cluster for usage frequency. The report is detailed, providing insights on image usage across different pods and CronJobs within namespaces and image usage frequency with and without version tags.

Usage

python3 runReport.py

Prerequisites

• kubectl must be installed and configured to connect to the Kubernetes cluster you intend to report on.
• The script requires a creds environment variable set to the path of the kubeconfig file used for cluster connection.

Output

The script will output a JSON file, image_cache.json, which contains the following properties:

• namespaces: An object where each key is a namespace name containing another object. This inner object uses pod/CronJob names as keys, and the values are arrays listing all images used by the containers within those pods/CronJobs.
• imageCount: An object keyed with raw image names (without version tags) and the number of their occurrences as values.
• imageWithTags: An object similar to imageCount, but includes version tags for each image with their occurrence counts as values.

Example image_cache.json snippet:

{
  "namespaces": {
    "default": {
      "pod1": [
        "nginx:1.17.1",
        "redis:6.0.5"
      ],
      ...
    },
    ...
  },
  "imageCount": {
    "nginx": 1,
    ...
  },
  "imageWithTags": {
    "nginx:1.17.1": 1,
    ...
  }
}

Notes

• The image_cache.json file is created in the directory where the script is executed.
• If an image_cache.json file already exists, it will be overwritten.
• It is recommended to rename the image_cache.json file post-execution to reflect the content or date of the report for better traceability.

runK8sImageReport.py

# Script Written By: Ariel Lothlorien
# Date: 2021-08-26
# Updated: 2021-11-02
# Purpose: To get a list of all images used in a Kubernetes cluster and how many times they are used.
#          The report is generated with details for each namespace, as well as counts for each image
#          both with and without their respective tags.
# Usage: python3 runReport.py
# Notes: This script requires kubectl to be installed and configured to connect to the cluster you want to report on.
#        This script will create a file called image_cache.json in the same directory as this script.
#        This script will overwrite the image_cache.json file if it already exists.
#        After use it is recommended to rename the image_cache.json file to something more descriptive.
#        The report output is a JSON file with 'namespaces', 'imageCount', and 'imageWithTags' properties.
#        'namespaces' is an object with keys as namespace names, each containing objects with pod/CronJob names as keys
#        and arrays of images as their values. 'imageCount' and 'imageWithTags' are objects keyed by the image names/tags
#        with the count of their occurrences as values.
#        This script will not work unless a creds variable is set to the path of a kubeconfig file to be used.

import json
import subprocess

# creds = '/home/username/.kube/config.dev'
# creds = '/home/username/.kube/config.stage'
# creds = '/home/username/.kube/config.prod'

# create cache for image counting
image_cache = {
    "namespaces": {
        # Example
        # podname: [image1, image2, image3]
    },
    "imageCount":{
        # Example
        # image: 2
    },
    "imageWithTags":{
        # Example
        # "org/image:tag": 4
    }
}

def get_pods_info():
    try:
        # Execute kubectl command to get all pods in all namespaces in JSON format
        result = subprocess.check_output(['kubectl', '--kubeconfig', creds, 'get', 'pods', '--all-namespaces', '-o', 'json'])
        results2 = subprocess.check_output(['kubectl', '--kubeconfig', creds, 'get', 'CronJobs', '--all-namespaces', '-o', 'json'])
        
        # Load JSON data from the result
        data = json.loads(result)
        data2 = json.loads(results2)
        # append results2['items'] to results['items']
        data['items'].extend(data2['items'])
        
        # Iterate over each pod and extract the required information
        for item in data['items']:
            namespace = item['metadata']['namespace']
            pod_name = item['metadata']['name']
            # if spec.containers is empty or does not exist, loop through them
            containers = item['spec'].get('containers', [])
            # if spec.jobTemplate.spec.template.spec.containers exists then use that
            if 'jobTemplate' in item['spec']:
                if 'spec' in item['spec']['jobTemplate']:
                    if 'template' in item['spec']['jobTemplate']['spec']:
                        if 'spec' in item['spec']['jobTemplate']['spec']['template']:
                            containers = item['spec']['jobTemplate']['spec']['template']['spec'].get('containers', [])
            
            for container in containers:
                image = container['image']
                print(f"Namespace: {namespace}, Pod: {pod_name}, Image: {image}")
                # Add image to cache
                if namespace not in image_cache['namespaces']:
                    image_cache['namespaces'][namespace] = {}
                if pod_name not in image_cache['namespaces'][namespace]:
                    image_cache['namespaces'][namespace][pod_name] = []
                image_cache['namespaces'][namespace][pod_name].append(image)

                # Split image and tag
                image_parts = image.split(':')
                image_name = image_parts[0]
                image_tag = image_parts[1] if len(image_parts) > 1 else 'latest'
                image_with_tag = f"{image_name}:{image_tag}"

                # Add image count to cache
                if image_name not in image_cache['imageCount']:
                    image_cache['imageCount'][image_name] = 0
                image_cache['imageCount'][image_name] += 1
                # Add image with tag count to cache
                if image_with_tag not in image_cache['imageWithTags']:
                    image_cache['imageWithTags'][image_with_tag] = 0
                image_cache['imageWithTags'][image_with_tag] += 1
    
    except subprocess.CalledProcessError as e:
        print(f"Error executing kubectl: {e}")
    except Exception as e:
        print(f"Unexpected error: {e}")

if __name__ == '__main__':
    get_pods_info()

    # Write cache to file
    with open('image_cache.json', 'w') as outfile:
        json.dump(image_cache, outfile, indent=4)