Priority Classes in Kubernetes: Scheduling

What Are Priority Classes?

A PriorityClass in Kubernetes is a non-namespaced object that assigns an integer priority value to pods. The higher the integer, the higher the scheduling priority.

They are used in:

• Pod scheduling: Determines which pods get scheduled first when resources are limited.

• Pod preemption: Allows high-priority pods to evict lower-priority pods if necessary.

Why Use Priority Classes?

Without priority classes, Kubernetes uses a default scheduling behavior that treats all pods equally. In real-world environments, this doesn’t work well—your logging daemon or critical API service should not be evicted in favor of a batch analytics job.

Priority Classes help by:

• Protecting critical workloads from being preempted.

• Allowing cluster resources to be allocated based on workload importance.

• Ensuring predictable behavior under resource pressure.

How to Create a Priority Class

Here’s a simple example of a PriorityClass definition:

apiVersion: scheduling.k8s.io/v1
kind: PriorityClass
metadata:
  name: high-priority
value: 100000
globalDefault: false
description: "This priority class is for critical production workloads."
# leaderElect (optional and advanced concept)

• value: Integer indicating the priority level (higher = more important).

• globalDefault: If true, pods without a specified priority class will use this.

• description: Useful for documentation.

Using Priority Class in a Pod

Once the class is defined, reference it in your pod spec:

apiVersion: v1
kind: Pod
metadata:
  name: my-critical-app
spec:
  priorityClassName: high-priority
  containers:
  - name: app
    image: nginx

Kubernetes will now schedule this pod with higher importance.

Pod Preemption: The Fine Print

If a high-priority pod cannot be scheduled due to lack of resources, Kubernetes can preempt (i.e., evict) lower-priority pods to make room. Some caveats:

• Only pods without PodDisruptionBudgets are considered for preemption.

• Preemption favors evicting the lowest-priority pods first.

• You can disable preemption at the pod level using:

preemptionPolicy: Never
# preemptionPolicy: PreemptLowerPriority # will evict lower priority pods

Example: Multiple Priority Classes

# Low priority for batch jobs
apiVersion: scheduling.k8s.io/v1
kind: PriorityClass
metadata:
  name: low-priority
value: 1000
globalDefault: false
---
# Medium priority for general services
apiVersion: scheduling.k8s.io/v1
kind: PriorityClass
metadata:
  name: medium-priority
value: 50000
globalDefault: true
---
# High priority for critical services
apiVersion: scheduling.k8s.io/v1
kind: PriorityClass
metadata:
  name: high-priority
value: 100000

How to View Priority Classes

You can list all defined PriorityClasses using:

kubectl get priorityclass

To describe a specific one:

kubectl describe priorityclass high-priority

Further Reading:

• Kubernetes Official Docs – Pod Priority and Preemption

• Quality of Service (QoS) in Kubernetes