Spread Workloads Across Zones via Topology Constraints for High Availability - Container Compute Service

When a zone becomes unavailable, workloads concentrated in that zone go offline entirely. Topology spread constraints distribute Pods evenly across availability zones in an ACS cluster so that a single-zone failure affects only the Pods in that zone — the rest of your workload keeps running.

Prerequisites

Before you begin, ensure that you have:

kube-scheduler installed, with a version that meets the following requirements:

ACS cluster version	Scheduler component version
1.31	v1.31.0-aliyun-1.2.0 and later
1.30	v1.30.3-aliyun-1.1.1 and later
1.28	v1.28.9-aliyun-1.1.0 and later

acs-virtual-node installed, version v2.12.0-acs.4 or later

Limitations

The following constraints apply only to Pods that meet all of these conditions:

The Pod uses the High-Performance Network GPU (GPU-HPN) compute type.
The Pod's schedulerName is default-scheduler.
Enable Custom Tags And Scheduler For GPU-HPN Nodes is not selected in the scheduler component configuration.

Note

New versions of the kube-scheduler component enable Enable Custom Tags And Scheduler For GPU-HPN Nodes by default. For details, see kube-scheduler.

For Pods that meet all three conditions, the following topology spread constraint fields behave differently from standard Kubernetes:

Field	Description	Constraint
`labelSelector`	Selects Pods to include when counting the number of Pods in each topology domain.	Pods of other compute types (General-purpose, compute-optimized, GPU) are excluded from the count.
`matchLabelKeys`	A list of label keys used together with `labelSelector` to identify the set of Pods for distribution calculation.	No constraint.
`nodeAffinityPolicy`	Controls how `nodeAffinity` and `nodeSelector` are applied when calculating topology distribution skew.	Not supported.
`nodeTaintsPolicy`	Controls how node taints are applied when calculating topology distribution skew.	Not supported.

For general-purpose, compute-optimized, and GPU compute types, the standard Kubernetes topology spread constraints behavior applies without these restrictions.

Distribute Pods across zones

View the virtual nodes in the cluster.

kubectl get node

The output lists nodes by zone. For example:

NAME                            STATUS   ROLES   AGE     VERSION
virtual-kubelet-cn-hangzhou-i   Ready    agent   5h42m   v1.28.3-xx
virtual-kubelet-cn-hangzhou-j   Ready    agent   5h42m   v1.28.3-xx

Create the dep-spread-demo.yaml file with the following content.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: dep-spread-demo
  labels:
    app: spread-demo
spec:
  replicas: 4
  selector:
    matchLabels:
      app: spread-demo
  template:
    metadata:
      labels:
        app: spread-demo
    spec:
      containers:
      - name: spread-demo
        image: registry-cn-beijing.ack.aliyuncs.com/acs/stress:v1.0.4
        command:
        - "sleep"
        - "infinity"
      # Spread Pods evenly across zones.
      # maxSkew: 1 means no zone can have more than one extra Pod compared to any other zone.
      topologySpreadConstraints:
      - maxSkew: 1
        topologyKey: topology.kubernetes.io/zone
        whenUnsatisfiable: DoNotSchedule
        labelSelector:
          matchLabels:
            app: spread-demo

Deploy the workload.
```
kubectl apply -f dep-spread-demo.yaml
```

Verify the Pod distribution across zones.

kubectl get pod -o wide

The output shows 4 Pods distributed across 2 zones, with 2 Pods per zone:

NAME                               READY   STATUS    RESTARTS   AGE     IP               NODE                            NOMINATED NODE   READINESS GATES
dep-spread-demo-7c656dbf5f-6twkc   1/1     Running   0          2m29s   192.168.xx.xxx   virtual-kubelet-cn-hangzhou-i   <none>           <none>
dep-spread-demo-7c656dbf5f-cgxr8   1/1     Running   0          2m29s   192.168.xx.xxx   virtual-kubelet-cn-hangzhou-j   <none>           <none>
dep-spread-demo-7c656dbf5f-f4fz9   1/1     Running   0          2m29s   192.168.xx.xxx   virtual-kubelet-cn-hangzhou-j   <none>           <none>
dep-spread-demo-7c656dbf5f-kc6xf   1/1     Running   0          2m29s   192.168.xx.xxx   virtual-kubelet-cn-hangzhou-i   <none>           <none>