ClusterSQL

Welcome to the ClusterSQL project! This project aims to provide a convenient way to access Kubernetes objects through a MySQL interface. The controller continuously watches for changes to Kubernetes objects and syncs them with in-memory MySQL, allowing users to use familiar SQL commands to interact with their Kubernetes clusters. Whether you are a MySQL expert looking for an easy way to slice and dice your Kubernetes resources, or just want a new way to visualize with your cluster, this project has something to offer.

This project was possible thanks to go-mysql-server project.

Dependencies

• Istio
• Kubernetes Metrics Server
• Prometheus

Scope of this project

• Be a convenient SQL-based component on observability stacks to slice and dice cluster informations
• Provide access to the most common objects (Pods, Containers, Nodes, Metrics, Affinites)

Non-goals:

• Support all K8s objects
• Edit K8s cluster through SQL

Build

docker build -t <image> -f ./build/Dockerfile .

Deployment

kubectl apply -f ./deployments

Limitations

ClusterSQL is a read-only interface. Any write query will not change the state of the monitored cluster, though the in-memory tables will be rewritten (TO-FIX).

The SQL limitations are listed in here

Data Model

Usage

Install mariadb-shell:

[fedora] $ sudo dnf install mariadb-shell

Expose port:

kubectl -n kube-system svc/clustersql 3306:3306

Connect and play:

mysql --host=0.0.0.0 --port=3306 --user=root kubernetes --execute="SELECT * FROM pods";