What are Kubernetes Topology Spread Constraints?

Question

Accepted Answer

**Topology Spread Constraints** — controlling Pod distribution across zones/nodes.

**Problem:** By default, scheduler may place all Pods on a single node/zone.

**Even distribution across zones:**
```yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: myapp
spec:
  replicas: 6
  template:
    spec:
      topologySpreadConstraints:
        - maxSkew: 1
          topologyKey: topology.kubernetes.io/zone
          whenUnsatisfiable: DoNotSchedule
          labelSelector:
            matchLabels:
              app: myapp
        - maxSkew: 1
          topologyKey: kubernetes.io/hostname
          whenUnsatisfiable: ScheduleAnyway
          labelSelector:
            matchLabels:
              app: myapp
```

**Parameters:**
- **maxSkew** — maximum difference in Pod count between topologies.
- **topologyKey** — key to distribute by (zone, hostname, label).
- **whenUnsatisfiable** — what to do if not satisfied.
  - `DoNotSchedule` — do not schedule.
  - `ScheduleAnyway` — schedule, but with priority.

**Distribution across nodes (anti-affinity alternative):**
```yaml
topologySpreadConstraints:
  - maxSkew: 1
    topologyKey: kubernetes.io/hostname
    whenUnsatisfiable: DoNotSchedule
    labelSelector:
      matchLabels:
        app: myapp
    matchLabelKeys:
      - pod-template-hash
```

**Topology Spread vs Pod Anti-Affinity:** Topology Spread is more flexible and predictable.

DevOps: What are Kubernetes Topology Spread Constraints?