content/operate/kubernetes/recommendations/node-selection.md
Kubernetes clusters often include nodes with different CPU and memory profiles. You control where Redis Enterprise cluster (REC) pods run by setting fields in the REC custom resource (CRD).
A Redis Enterprise cluster (REC) runs as a StatefulSet. The Kubernetes scheduler assigns nodes when you create or resize the cluster, or when a pod restarts.
Use these options to control pod placement:
The [nodeSelector]({{<relref "/operate/kubernetes/reference/api/redis_enterprise_cluster_api#spec">}}) field matches the Kubernetes nodeSelector syntax.
Label the nodes you want to target. For example, if nodes 'n1' and 'n2' are labeled with memory=high:
kubectl label nodes n1 memory=high
kubectl label nodes n2 memory=high
The Redis Enterprise cluster CRD can request to be scheduled on these nodes:
apiVersion: app.redislabs.com/v1
kind: RedisEnterpriseCluster
metadata:
name: rec
spec:
nodes: 3
nodeSelector:
memory: high
The operator copies [nodeSelector]({{< relref "/operate/kubernetes/reference/api/redis_enterprise_cluster_api#spec" >}}) into the pod spec. The scheduler places pods only on nodes that match the selector.
Node pools group similar nodes. Providers label nodes by pool.
Use [nodeSelector]({{< relref "/operate/kubernetes/reference/api/redis_enterprise_cluster_api#spec" >}}) to target a pool by label. For example, on GKE:
apiVersion: app.redislabs.com/v1
kind: RedisEnterpriseCluster
metadata:
name: rec
spec:
nodes: 3
nodeSelector:
cloud.google.com/gke-nodepool: 'high-memory'
Cloud providers label nodes by pool. See links below for specific documentation.
Use node taints and pod tolerations to control REC pod scheduling. Set tolerations with [spec.podTolerations]({{< relref "/operate/kubernetes/reference/api/redis_enterprise_cluster_api#specpodtolerations" >}}) (standard Kubernetes tolerations).
Example: on a single node pool, reserve nodes n1–n3 for REC by adding taints:
kubectl taint nodes n1 db=rec:NoSchedule
kubectl taint nodes n2 db=rec:NoSchedule
kubectl taint nodes n3 db=rec:NoSchedule
This blocks pods unless they tolerate the db=rec taint.
Then add a matching toleration to the REC:
apiVersion: app.redislabs.com/v1
kind: RedisEnterpriseCluster
metadata:
name: rec
spec:
nodes: 3
podTolerations:
- key: db
operator: Equal
value: rec
effect: NoSchedule
A set of taints can also handle more complex use cases.
For example, a role=test or role=dev taint can be used to designate a node as dedicated for testing or development workloads via pod tolerations.
By default, REC node pods are not scheduled on the same node within the same cluster:
podAntiAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchLabels:
app: redis-enterprise
redis.io/cluster: rec
redis.io/role: node
topologyKey: kubernetes.io/hostname
Each pod has these labels. redis.io/cluster is your cluster name.
Modify this rule to widen or narrow placement. For example, remove the redis.io/cluster label to prevent pods from different clusters from sharing a node:
apiVersion: app.redislabs.com/v1
kind: RedisEnterpriseCluster
metadata:
name: rec
spec:
nodes: 3
podAntiAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchLabels:
app: redis-enterprise
redis.io/role: node
topologyKey: kubernetes.io/hostname
To avoid co-locating with other database workloads, label those pods local/role: database and add anti-affinity to keep one database per node:
apiVersion: app.redislabs.com/v1
kind: RedisEnterpriseCluster
metadata:
name: rec
spec:
nodes: 3
extraLabels:
local/role: database
podAntiAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchLabels:
local/role: database
app: redis-enterprise
redis.io/cluster: rec
redis.io/role: node
topologyKey: kubernetes.io/hostname
Kubernetes will not schedule two pods with label local/role: database on the same node.
Enable rack-zone awareness to improve availability during rack or zone failures.
{{%note%}}When creating your rack-zone ID, there are some constraints to consider; see [rack-zone awareness]({{< relref "/operate/rs/clusters/configure/rack-zone-awareness#rack-zone-id-rules" >}}) for more info. {{%/note%}}
Configure it with [spec.rackAwarenessNodeLabel]({{< relref "/operate/kubernetes/reference/api/redis_enterprise_cluster_api#spec" >}}) in the REC.
The most common label used for rack-zone awareness is topology.kubernetes.io/zone, a standard Kubernetes label that shows the zone a node runs in. Many Kubernetes platforms add this label to nodes by default, as noted in the Kubernetes documentation.
If your platform doesn’t set this label automatically, you can use any custom label that describes the node’s topology (such as rack, zone, or region).
{{< warning >}} All eligible nodes must have the label for rack awareness to work. The operator requires every node that might run Redis Enterprise pods to be labeled. If any nodes are missing the label, reconciliation fails. {{< /warning >}}
Eligible nodes are nodes where REC pods can run. By default, this means all worker nodes. You can limit eligibility with [spec.nodeSelector]({{< relref "/operate/kubernetes/reference/api/redis_enterprise_cluster_api#spec" >}}).
Give each eligible node a label value that reflects its rack, zone, or region.
Check node label values:
kubectl get nodes -o custom-columns="name:metadata.name","rack\\zone:metadata.labels.topology\.kubernetes\.io/zone"
name rack\zone
ip-10-0-x-a.eu-central-1.compute.internal eu-central-1a
ip-10-0-x-b.eu-central-1.compute.internal eu-central-1a
ip-10-0-x-c.eu-central-1.compute.internal eu-central-1b
ip-10-0-x-d.eu-central-1.compute.internal eu-central-1b
Grant the operator read access to node labels with a ClusterRole and ClusterRoleBinding.
ClusterRole:
{{<embed-yaml "k8s/rack_aware_cluster_role.md" "rack-aware-cluster-role.yaml">}}
Bind to the redis-enterprise-operator service account:
{{<embed-yaml "k8s/rack_aware_cluster_role_binding.md" "rack-aware-cluster-role-binding.yaml">}}
Apply these files with kubectl apply. For example:
kubectl apply -f rack-aware-cluster-role.yaml
kubectl apply -f rack-aware-cluster-role-binding.yaml
After you apply the role and binding, you can configure rack awareness.
Set [spec.rackAwarenessNodeLabel]({{< relref "/operate/kubernetes/reference/api/redis_enterprise_cluster_api#spec" >}}) to the node label to use:
{{<embed-yaml "k8s/rack_aware_rec.md" "rack-aware-cluster.yaml">}}
{{< note >}}
When you set [spec.rackAwarenessNodeLabel]({{< relref "/operate/kubernetes/reference/api/redis_enterprise_cluster_api#spec" >}}), the operator sets the anti-affinity topologyKey to that label unless you define [spec.podAntiAffinity]({{< relref "/operate/kubernetes/reference/api/redis_enterprise_cluster_api#specpodantiaffinity" >}}). If you define both, make sure topologyKey matches your node label.
{{< /note >}}
{{< warning >}} Pod restart distribution maintenance: When rack awareness is enabled, node pods and shards are initially deployed based on rack constraints to ensure proper distribution across zones. However, Redis Enterprise does not automatically maintain this distribution when node pods are restarted.
After pod restarts, the rack awareness policy may be violated, requiring manual intervention to restore proper shard distribution. While Redis Enterprise provides tools to identify shards that need to be moved to restore correct rack distribution, it does not provide automated orchestration to perform these moves. {{< /warning >}}
Important considerations for production deployments: