Redis FAQs

1. How do I configure scheduling policies for a Redis Cluster?

You can use the schedulingPolicy field to control how Redis Pods are distributed across your Kubernetes nodes. This is useful for improving high availability, optimizing resource usage, or meeting specific placement requirements.

Below are common scheduling scenarios with example configurations.

Distribute Pods within the same shard across different nodes

To ensure high availability within a shard, use podAntiAffinity to prevent Pods of the same component/shard from running on the same node. Since each shard name are generated by KubeBlocks, there is no way set the value beforehand. You can use the placeholder "" and KubeBlocks will automatically replace it with the actual shard name.

schedulingPolicy:
  affinity:
    podAntiAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
        - labelSelector:
            matchLabels:
              apps.kubeblocks.io/component-name: ""  # Leave it blank, automatically replaced by KubeBlocks
          topologyKey: kubernetes.io/hostname

Distribute shards across different availability zones

To improve fault tolerance at the zone level, use podAntiAffinity to spread all shards within a sharding group across different availability zones.

schedulingPolicy:
  affinity:
    podAntiAffinity:
      preferredDuringSchedulingIgnoredDuringExecution:
        - weight: 100
          podAffinityTerm:
            labelSelector:
              matchLabels:
                apps.kubeblocks.io/sharding-name: my-sharding  # Replace with your sharding name
            topologyKey: topology.kubernetes.io/zone

Co-locate Pods with a specific cluster

To reduce network latency or optimize data locality, use podAffinity to schedule Pods on the same nodes as an existing cluster.

schedulingPolicy:
  affinity:
    podAffinity:
      preferredDuringSchedulingIgnoredDuringExecution:
        - weight: 50
          podAffinityTerm:
            labelSelector:
              matchLabels:
                app.kubernetes.io/instance: my-cluster  # Replace with the target cluster name
            topologyKey: kubernetes.io/hostname

Label selector reference

Use the following labels to target different scopes in your scheduling policies: