Skip to main content

​
Prerequisites

Before installing Redis Operator, ensure you have:
  • Kubernetes 1.25 or later
  • kubectl configured to access your cluster
  • Helm 3.8 or later (for Helm installation)
  • Cluster-admin permissions to install CRDs and RBAC resources
The operator requires cluster-wide permissions to manage RedisCluster resources across namespaces.

​
Installation Methods

1

Install CRDs

Install the Custom Resource Definitions for RedisCluster, RedisBackup, and RedisScheduledBackup:
kubectl apply -f https://raw.githubusercontent.com/howl-cloud/redis-operator/main/config/crd/bases/redis.io_redisclusters.yaml
kubectl apply -f https://raw.githubusercontent.com/howl-cloud/redis-operator/main/config/crd/bases/redis.io_redisbackups.yaml
kubectl apply -f https://raw.githubusercontent.com/howl-cloud/redis-operator/main/config/crd/bases/redis.io_redisscheduledbackups.yaml
Verify the CRDs are installed:
kubectl get crds | grep redis.io
Expected output:
redisbackups.redis.io                          2026-02-28T10:30:00Z
redisclusters.redis.io                         2026-02-28T10:30:00Z
redisscheduledbackups.redis.io                 2026-02-28T10:30:00Z
2

Install the Operator

Install the operator using Helm:
helm upgrade --install redis-operator oci://ghcr.io/howl-cloud/charts/redis-operator \
  --namespace redis-operator-system \
  --create-namespace \
  --version 0.1.0
The operator runs as a single replica with leader election enabled by default. For production, this provides high availability if the pod is rescheduled.
3

Verify Installation

Check that the operator is running:
kubectl get pods -n redis-operator-system
Expected output:
NAME                              READY   STATUS    RESTARTS   AGE
redis-operator-7d4b9c5f6b-xyz12   1/1     Running   0          30s
Check the operator logs:
kubectl logs -n redis-operator-system -l app.kubernetes.io/name=redis-operator

​
Helm Configuration

The Helm chart exposes the following configuration values from charts/redis-operator/Chart.yaml: Chart Metadata:
  • Version: 0.1.0
  • App Version: 0.1.0
  • Description: A Kubernetes operator for managing Redis 7.2 clusters with automatic failover, rolling updates, and backup support

​
Common Configuration Options

# Number of operator replicas (usually 1 with leader election)
replicaCount: 1

image:
  repository: redis-operator
  pullPolicy: IfNotPresent
  tag: ""  # Defaults to chart appVersion

resources:
  limits:
    cpu: 500m
    memory: 256Mi
  requests:
    cpu: 100m
    memory: 128Mi

leaderElection:
  enabled: true

metrics:
  enabled: true
  port: 9090

webhook:
  enabled: true
  port: 9443

​
Install with Custom Values

Create a values.yaml file with your configuration:
values.yaml
image:
  repository: ghcr.io/howl-cloud/redis-operator
  tag: "0.1.0"

resources:
  limits:
    cpu: 1000m
    memory: 512Mi
  requests:
    cpu: 200m
    memory: 256Mi

metrics:
  enabled: true
  serviceMonitor:
    enabled: true

monitoring:
  alertingRules:
    enabled: true
  grafanaDashboard:
    enabled: true
  podMonitor:
    enabled: true
Install with custom values: 
helm upgrade --install redis-operator oci://ghcr.io/howl-cloud/charts/redis-operator \
  --namespace redis-operator-system \
  --create-namespace \
  --values values.yaml

​
Monitoring Setup

​
Prometheus Metrics

The operator exposes Prometheus metrics on port 9090 by default:
  • Controller metrics: Reconciliation duration, queue depth, error rates
  • Instance metrics: Redis instance health, replication lag, memory usage
# Enable ServiceMonitor for operator controller metrics
metrics:
  enabled: true
  serviceMonitor:
    enabled: true
    interval: 30s
    scrapeTimeout: 10s
    labels:
      prometheus: kube-prometheus

​
Alerting Rules

Enable default Prometheus alerting rules:
values.yaml
monitoring:
  alertingRules:
    enabled: true
    # Replication lag threshold (10MB default)
    replicationLagThresholdBytes: 10485760
    # Memory usage threshold (85% default)
    memoryUsageThresholdRatio: 0.85
    # Backup age threshold (24 hours default)
    backupMissingSeconds: 86400
Default alerts include:
  • RedisReplicationLagHigh: Replica lag exceeds threshold
  • RedisMemoryUsageHigh: Memory usage above threshold
  • RedisBackupMissing: No successful backup in threshold period
  • RedisClusterDegraded: Cluster not in Healthy phase

​
Grafana Dashboard

Enable the default Grafana dashboard:
values.yaml
monitoring:
  grafanaDashboard:
    enabled: true
    labels:
      grafana_dashboard: "1"
The dashboard includes:
  • Cluster health and phase transitions
  • Replication lag and topology
  • Memory usage and eviction rates
  • Operations per second and latency
  • Backup success rates

​
Webhook Configuration

The operator includes admission webhooks for validation and defaulting. The PKI is self-managed and does not require cert-manager.
Webhooks are enabled by default:
values.yaml
webhook:
  enabled: true
  port: 9443
  # Leave empty to auto-generate and rotate certificates
  certSecretName: ""
The operator automatically:
  1. Generates a self-signed CA
  2. Issues a server certificate for the webhook service
  3. Rotates certificates before expiration
  4. Updates ValidatingWebhookConfiguration and MutatingWebhookConfiguration with the CA bundle

​
Disable Webhooks

Disabling webhooks removes validation and defaulting. Invalid configurations may be accepted and cause reconciliation failures.
To disable webhooks: 
helm upgrade --install redis-operator oci://ghcr.io/howl-cloud/charts/redis-operator \
  --namespace redis-operator-system \
  --create-namespace \
  --set webhook.enabled=false

​
RBAC and Security

The operator requires cluster-wide permissions defined in config/rbac/role.yaml:
  • Cluster-scoped: CRDs, ClusterRoles, ClusterRoleBindings
  • Namespace-scoped: Pods, PVCs, Services, Secrets, ConfigMaps, Events
  • Leader election: Leases in the operator namespace
All resources are created with:
Security Context (Pod)
podSecurityContext:
  runAsNonRoot: true
  seccompProfile:
    type: RuntimeDefault
Security Context (Container)
containerSecurityContext:
  allowPrivilegeEscalation: false
  readOnlyRootFilesystem: true
  runAsNonRoot: true
  capabilities:
    drop:
      - ALL

​
Upgrade the Operator

Upgrade to a new version: 
helm upgrade redis-operator oci://ghcr.io/howl-cloud/charts/redis-operator \
  --namespace redis-operator-system \
  --version 0.2.0
Upgrading the operator does not automatically upgrade existing Redis clusters. You must update the spec.imageName field in each RedisCluster resource to trigger a rolling update.

​
Uninstall

Uninstall the operator: 
helm uninstall redis-operator --namespace redis-operator-system
Uninstalling the operator does NOT delete existing RedisCluster resources or their data. The CRDs and custom resources remain in the cluster.
To completely remove the operator and CRDs: 
# Delete all RedisCluster resources first
kubectl delete redisclusters --all --all-namespaces

# Uninstall the operator
helm uninstall redis-operator --namespace redis-operator-system

# Delete CRDs (this deletes all custom resources)
kubectl delete crd redisclusters.redis.io
kubectl delete crd redisbackups.redis.io
kubectl delete crd redisscheduledbackups.redis.io

​
Next Steps

Quick Start

Deploy your first Redis cluster

API Reference

Explore RedisCluster configuration options

Build docs developers (and LLMs) love

Get started for freeTalk to us