What is Kubernetes Pod Scheduling?

Question

Accepted Answer

**Pod Scheduling** — process of assigning Pods to Nodes based on resource requirements and constraints.

```yaml
apiVersion: v1
kind: Pod
metadata:
  name: gpu-pod
spec:
  nodeSelector:
    accelerator: nvidia-tesla-v100
  affinity:
    podAffinity:
      preferredDuringSchedulingIgnoredDuringExecution:
      - weight: 100
        podAffinityTerm:
          labelSelector:
            matchLabels:
              app: cache
          topologyKey: kubernetes.io/hostname
    nodeAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
        nodeSelectorTerms:
        - matchExpressions:
          - key: zone
            operator: In
            values:
            - us-west-2a
  tolerations:
  - key: "dedicated"
    operator: "Equal"
    value: "gpu"
    effect: "NoSchedule"
  topologySpreadConstraints:
  - maxSkew: 1
    topologyKey: topology.kubernetes.io/zone
    whenUnsatisfiable: DoNotSchedule
    labelSelector:
      matchLabels:
        app: myapp
```

**Scheduling Plugins:**
- nodeSelector: simple node selection.
- affinity/anti-affinity: complex placement rules.
- tolerations: allow scheduling on tainted nodes.
- topologySpreadConstraints: distribute Pods evenly.
- priority: preemption of lower-priority Pods.

DevOps: What is Kubernetes Pod Scheduling?