Priority classes for critical workloads in Kubernetes - Step-by-Step Execution

Process Flow - Priority classes for critical workloads

Define PriorityClass

↓

Assign PriorityClass to Pod

↓

Scheduler checks Pod priority

↓

If resources scarce

Yes No↓

Evict lower priority Pods

↓

Critical workload runs

This flow shows how defining a PriorityClass and assigning it to a Pod affects scheduling and eviction decisions for critical workloads.

Execution Sample

Kubernetes

apiVersion: scheduling.k8s.io/v1
kind: PriorityClass
metadata:
  name: high-priority
value: 1000000
---
apiVersion: v1
kind: Pod
metadata:
  name: critical-pod
spec:
  priorityClassName: high-priority
  containers:
  - name: app
    image: nginx

Defines a high priority class and assigns it to a critical Pod to ensure it is scheduled before lower priority Pods.

Process Table

Step	Action	Pod Priority	Scheduler Decision	Result
1	Define PriorityClass 'high-priority'	-	-	PriorityClass created with value 1000000
2	Create Pod with priorityClassName 'high-priority'	1000000	-	Pod marked with high priority
3	Scheduler checks available resources	-	-	Resources are limited
4	Scheduler compares Pod priority with running Pods	1000000 vs 1000	Pod priority higher	Lower priority Pod evicted
5	Scheduler schedules critical Pod	1000000	Scheduled	Critical Pod runs
6	No more Pods to schedule	-	-	Scheduling complete

💡 All Pods scheduled or evicted based on priority; critical Pod runs successfully

Status Tracker

Variable	Start	After Step 2	After Step 4	Final
PriorityClass 'high-priority' value	undefined	1000000	1000000	1000000
Pod 'critical-pod' priority	undefined	1000000	1000000	1000000
Running Pods count	5 (mixed priorities)	5	4 (one low priority evicted)	4

Key Moments - 2 Insights

Why does the scheduler evict a running Pod when a new Pod with higher priority is created?

The scheduler evicts lower priority Pods to free resources for higher priority Pods, as shown in step 4 where the Pod with priority 1000000 causes eviction of a Pod with priority 1000.

What happens if two Pods have the same priority?

If priorities are equal, the scheduler uses other factors like creation time or resource requests to decide scheduling, but priority classes mainly affect eviction order as seen in the execution table.

Visual Quiz - 3 Questions

Look at the execution table, what is the priority value assigned to the critical Pod at step 2?

A 1000

B 1000000

C 0

D Undefined

💡 Hint

Check the 'Pod Priority' column at step 2 in the execution table.

At which step does the scheduler evict a lower priority Pod?

A Step 3

B Step 5

C Step 4

D Step 6

💡 Hint

Look for the 'Scheduler Decision' column mentioning eviction in the execution table.

If the critical Pod had a lower priority than running Pods, what would happen?

A It might be pending or evicted itself

B It would evict higher priority Pods

C It would be scheduled immediately

D Scheduler ignores priority

💡 Hint

Refer to the scheduler behavior in steps 3 and 4 about resource scarcity and eviction.

Concept Snapshot

PriorityClass defines importance of Pods.
Higher priority Pods get scheduled first.
Lower priority Pods can be evicted to free resources.
Assign PriorityClass to Pod via priorityClassName.
Critical workloads use high priority to ensure running.

Full Transcript

Priority classes in Kubernetes help the scheduler decide which Pods are more important. You create a PriorityClass with a numeric value. Then, assign it to a Pod using priorityClassName. When resources are tight, the scheduler evicts lower priority Pods to make room for higher priority ones. This ensures critical workloads run reliably. The execution flow shows defining a PriorityClass, creating a Pod with it, scheduler checking resources, evicting lower priority Pods, and scheduling the critical Pod.