Skip to main content

Chapter 2.5 - Scheduler

Learning Objectives​

By the end of this chapter, you will be able to:

  • Understand the role of the Scheduler
  • Explain the scheduling process
  • Understand filters and scores
  • Identify scheduling constraints

What is the Scheduler?​

The Scheduler (kube-scheduler) is responsible for assigning unscheduled Pods to appropriate nodes in the cluster. It takes into account resource constraints, affinities, and policies.

Scheduling Process​

Scheduling Steps​

1. Filtering​

Eliminates nodes that cannot run the Pod:

  • Predicate 1: Sufficient resources (CPU, memory)
  • Predicate 2: Affinities/anti-affinities respected
  • Predicate 3: Taints and tolerations
  • Predicate 4: Node selectors

2. Scoring (Prioritization)​

Scores each eligible node (0-10):

  • Priority 1: Resource balancing
  • Priority 2: Affinity with other Pods
  • Priority 3: Data locality
  • Priority 4: User preferences

3. Selection​

The node with the highest score is selected.

Scheduling Example​

Scenario​

apiVersion: v1
kind: Pod
metadata:
  name: my-pod
spec:
  containers:
  - name: app
    image: nginx
    resources:
      requests:
        cpu: "500m"
        memory: "512Mi"

Process​

Scheduling Constraints​

1. Resources (Requests/Limits)​

resources:
  requests:
    cpu: "500m"
    memory: "512Mi"
  limits:
    cpu: "1000m"
    memory: "1Gi"

The Scheduler verifies that the node has enough available resources.

2. Node Selectors​

spec:
  nodeSelector:
    disktype: ssd
    zone: us-east-1a

The Pod will only be scheduled on nodes with these labels.

3. Affinities (Affinity)​

affinity:
  nodeAffinity:
    requiredDuringSchedulingIgnoredDuringExecution:
      nodeSelectorTerms:
      - matchExpressions:
        - key: zone
          operator: In
          values:
          - us-east-1a

4. Taints and Tolerations​

# Taint on the node
kubectl taint nodes node1 key=value:NoSchedule

# Toleration in the Pod
tolerations:
- key: "key"
  operator: "Equal"
  value: "value"
  effect: "NoSchedule"

Scheduling Strategies​

Resource Balancing​

The Scheduler tries to balance load across nodes:

Locality Preferences​

The Scheduler prefers scheduling Pods near their data (zone affinity).

Custom Scheduler​

You can create your own scheduler for specific needs:

apiVersion: v1
kind: Pod
spec:
  schedulerName: my-custom-scheduler
  containers:
  - name: app
    image: nginx

Useful Commands​

# View Scheduler logs
kubectl logs -n kube-system kube-scheduler-<node>

# See why a Pod is not scheduled
kubectl describe pod <pod-name>

# View scheduling events
kubectl get events --field-selector involvedObject.name=<pod-name>

Summary​

In this chapter, you learned:

Scheduler: Assigns Pods to appropriate nodes
Process: Filtering -> Scoring -> Selection -> Bind
Constraints: Resources, selectors, affinities, taints
Balancing: Optimal load distribution
Customization: Custom schedulers are possible

Next Steps​

Now that you understand the Scheduler:

Chapter 2.6: kubelet - Worker Node Agent
Chapter 2.7: kube-proxy

Chapter created: December 2024