⚠️ Taints and Tolerations in Kubernetes

Taints prevent Pods from being scheduled onto certain nodes unless those Pods tolerate the taint. It’s a way to repel unwanted workloads.

Tolerations are the Pod’s way of saying, “I’m okay with that taint.”

---

✅ Why Use Taints & Tolerations?

• Dedicate nodes for specific workloads (e.g., GPU jobs, system services)
• Prevent scheduling critical apps on generic nodes
• Enforce workload separation

---

🧱 How Taints Work

Taints are applied to nodes:

kubectl taint nodes <node-name> key=value:effect

Effects:

Effect	Description
NoSchedule	Pod won’t be scheduled unless it tolerates taint
PreferNoSchedule	Tries to avoid, but not guaranteed
NoExecute	Evicts existing pods without toleration

---

🔧 Example: Apply Taint

kubectl taint nodes node1 dedicated=gpu:NoSchedule

This means: Only Pods with a matching toleration can run on node1.

---

🩹 Example: Pod With Toleration

apiVersion: v1
kind: Pod
metadata:
  name: gpu-pod
spec:
  containers:
  - name: gpu-app
    image: busybox
    command: ["sleep", "3600"]
  tolerations:
  - key: "dedicated"
    operator: "Equal"
    value: "gpu"
    effect: "NoSchedule"

---

📤 Remove Taint

kubectl taint nodes node1 dedicated=gpu:NoSchedule-

The - at the end removes the taint.

---

🤔 Use Cases

Use Case	Taint Command Example
Reserve nodes for GPU apps	dedicated=gpu:NoSchedule
Evict Pods on node shutdown	key=value:NoExecute
Prefer node for a workload	key=value:PreferNoSchedule

---

🔍 View Node Taints

kubectl describe node <node-name>

Look for Taints: section in the output.

---

✅ Summary

Concept	Applied To	Purpose
Taint	Node	Repel unwanted Pods
Toleration	Pod	Allows Pod to schedule on tainted node

• Taints repel pods
• Tolerations allow pods to be scheduled anyway
• Common for dedicated infrastructure, node pools, or eviction control