Resource quotas per namespace in Kubernetes - Step-by-Step Execution

Process Flow - Resource quotas per namespace

Create Namespace

↓

Define ResourceQuota YAML

↓

Apply ResourceQuota to Namespace

↓

Kubernetes Enforces Limits

↓

Pods/Resources Created Within Quota

↓

Quota Usage Updated

↓

Reject Excess Resource Requests

This flow shows how a resource quota is created and applied to a namespace, then Kubernetes enforces resource limits on pods and other objects within that namespace.

Execution Sample

Kubernetes

apiVersion: v1
kind: ResourceQuota
metadata:
  name: compute-quota
  namespace: dev
spec:
  hard:
    pods: "3"
    requests.cpu: "2"
    requests.memory: "4Gi"

This YAML defines a resource quota named 'compute-quota' in the 'dev' namespace limiting pods to 3, CPU requests to 2 cores, and memory requests to 4Gi.

Process Table

Step	Action	Namespace State	Quota Usage	Result
1	Create namespace 'dev'	Namespace 'dev' created	0 pods, 0 CPU, 0 Memory	Namespace ready
2	Apply ResourceQuota 'compute-quota'	Quota set: pods=3, CPU=2, Memory=4Gi	0 pods, 0 CPU, 0 Memory	Quota active
3	Create Pod1 requesting 1 CPU, 1Gi memory	Pods=1	CPU=1, Memory=1Gi	Pod1 created
4	Create Pod2 requesting 1 CPU, 2Gi memory	Pods=2	CPU=2, Memory=3Gi	Pod2 created
5	Create Pod3 requesting 1 CPU, 1Gi memory	Pods=2	CPU=2, Memory=3Gi	Rejected: CPU quota exceeded
6	Delete Pod1	Pods=1	CPU=1, Memory=2Gi	Pod1 deleted, quota freed
7	Create Pod3 requesting 1 CPU, 1Gi memory	Pods=2	CPU=2, Memory=3Gi	Pod3 created
8	Create Pod4 requesting 1 CPU, 3Gi memory	Pods=2	CPU=2, Memory=3Gi	Rejected: Memory quota exceeded
9	Create Pod4 requesting 1 CPU, 1Gi memory	Pods=2	CPU=2, Memory=3Gi	Rejected: CPU quota exceeded

💡 Pods quota limit of 3 and CPU/memory limits enforced, excess requests rejected

Status Tracker

Variable	Start	After Step 3	After Step 4	After Step 5	After Step 6	After Step 7	After Step 8	After Step 9
Pods Count	0	1	2	2	1	2	2	2
CPU Requests (cores)	0	1	2	2	1	2	2	2
Memory Requests (Gi)	0	1	3	3	2	3	3	3

Key Moments - 3 Insights

Why was Pod3 creation rejected at step 5 even though pods quota was not exceeded?

At step 5, CPU requests would exceed the quota of 2 cores (would become 3), so Kubernetes rejected Pod3. See execution_table row 5.

Why can Pod3 be created after deleting Pod1 at step 7?

Deleting Pod1 frees CPU and memory quota, so Pod3 fits within limits now. See execution_table rows 6 and 7.

Why was Pod4 rejected at step 9 despite requesting only 1 CPU and 1Gi memory?

CPU quota is 2 cores total. After step 7 (and 8), 2 CPU cores are in use by Pod2 and Pod3. Adding Pod4's 1 CPU exceeds it. See execution_table row 9.

Visual Quiz - 3 Questions

Look at the execution table, what is the CPU requests usage after step 4?

A 1 core

B 2 cores

C 3 cores

D 0 cores

💡 Hint

Check the 'Quota Usage' column at step 4 in the execution_table.

At which step does the first pod rejection occur?

A Step 5

B Step 7

C Step 3

D Step 9

💡 Hint

Check the 'Result' column in the execution_table for the first 'Rejected'.

If Pod2 requested only 0.5 CPU instead of 1 CPU, what would happen at step 5?

A Pod3 creation would still be rejected

B Pod2 creation would be rejected

C Pod3 creation would succeed

D Pods quota would be exceeded

💡 Hint

Consider CPU usage after step 4 and how it affects step 5 in execution_table.

Concept Snapshot

Resource Quotas limit resource use per namespace in Kubernetes.
Define a ResourceQuota YAML with 'hard' limits.
Apply it to a namespace to enforce limits on pods, CPU, memory.
Kubernetes rejects resource requests exceeding quota.
Quota usage updates dynamically as pods are created/deleted.

Full Transcript

Resource quotas in Kubernetes control how many resources like pods, CPU, and memory can be used in a namespace. First, you create a namespace, then define a ResourceQuota YAML specifying limits. When you apply this quota to the namespace, Kubernetes tracks resource usage. If you try to create pods or request resources beyond the quota, Kubernetes rejects those requests. Deleting pods frees quota, allowing new pods within limits. This ensures fair resource sharing and prevents one namespace from using too much.