How to Build a Custom Kubernetes Scheduler Using Simple Bash Scripts

A hands-on lab guide demonstrates how Kubernetes scheduling works by walking through the creation of a custom scheduler from scratch using bash scripts. The Kubernetes scheduler follows three phases — filtering, scoring, and binding — to intelligently assign pods to nodes based on CPU, memory, and other constraints. The lab shows how to configure an nginx pod to use a custom scheduler named 'my-scheduler,' which causes the pod to remain in a Pending state until that scheduler is active in the cluster. A simple open-source bash script is then used to mimic core scheduler behavior, including querying available nodes, identifying waiting pods, and making binding requests to the Kubernetes API server. While production schedulers are typically built in Golang, this approach offers a clear, accessible way for DevOps engineers to understand the underlying scheduling workflow.