namespace-usage

Namespace Usage

Ouputs a JSON report showing

• the CPU requested vs. used, for all namespaces
• the Memory requested vs. used for all namespaces
• the number of Hard limits set for pods vs current number of running pods for all namespaces
• number of containers for all namespaces

The main package in this report will perform the following steps:

• fetch the kubeconfig from the s3 bucket
• authenticate to the kubernetes cluster and set the current context to ctx env variable
• get all namespaces
• get all pods and create a resource requests Map of NamespaceResource type
• get all pod metrics and create a resource usage map of NamespaceResource type
• get all resourcequota from cluster and create a quota map
• build a usageReport with all the data required i.e cpu, memory and pods
• post them as json to the namespace_usage endpoint

Environment variables

You can see from the codebase, a number of environment variables are required to run the program. These are:

• bucket - The bucket name that hosts a kubeconfig file, commonly used in cloud-platform.

• context - The kubernetes cluster to which the report

• hoodawAPIKey: The API key of the "How out of date are we application" (HOODAW)

• hoodawEndpoint: The endpoint of the HOODAW Application "namespace_usage"

• hoodawHost: The hostname of the HOODAW hostname i.e. https://reports.cloud-platform.service.justice.gov.uk

• kubeconfig - The kubeconfig name in the variable bucket

• region - AWS Region to get the s3 bucket

• kubeCfgPath - Path in which the kubeconfig has to be stored

How to test locally

From the root of the HOODAW directory, run make dev-server. Ensure your environment variables are set i.e. hoodawHost=http://localhost:4567.

Either run go run main.go with arguments specified, or simply run go test -v ..