repository·main·Indexed 23 days ago
https://github.com/ot-container-kit/redis-operatorA Golang-based Kubernetes operator designed to automate the deployment and management of Redis in Standalone, Cluster, Replication, and Sentinel modes. It provides Helm charts for the operator itself and for deploying Redis clusters and replications, featuring support for cert-manager webhook integration, TLS configuration, and production best practices for memory management and persistence.
Redis Operator is a Golang-based Kubernetes operator designed to automate the deployment and management of Redis instances. It supports multiple deployment modes including Standalone, Cluster, Replication, and Sentinel, ensuring best practices for both Cloud and Bare-metal environments.
Supported Redis Modes:
Important Version Constraint:
This operator only supports Redis versions =>6.
The Redis Operator is a Golang-based Kubernetes operator designed to manage Redis workloads using production-ready best practices. It supports multiple deployment modes, including:
Key Security and Networking Features:
redis-exporter.The Redis operator supports four primary deployment strategies for managing Redis instances:
Redis Sentinel is a tool that provides automatic failover and monitoring for Redis nodes. It runs separate processes that communicate with each other and with Redis nodes to:
It can also perform additional tasks such as sending notifications and managing configuration changes, making it a robust solution for implementing high availability in Redis.
A Redis cluster uses a data sharding strategy to automatically partition data across multiple Redis nodes, providing distributed storage and preventing a single point of failure.
In this architecture, if a Redis node fails, a follower pod is automatically promoted to a leader. When the failed node returns online, it resumes its role as a follower.
A sharded cluster with a leader-only architecture requires a minimum of 3 nodes. If followers are included for high availability, the deployment will consist of at least 6 pods/processes.
The Redis Operator codebase is organized into several key directories:
api/: Contains the interface and structure for Custom Resource Definition (CRD) definitions (e.g., v1beta2).controllers/: Contains the watch controllers responsible for creating, updating, and deleting resources.k8sutils/: Contains the codebase for Kubernetes resources such as StatefulSets and Services.config/: Contains configuration for certmanager, crd (bases and patches), rbac, prometheus, and samples.bin/: Binary files.hack/: Internal scripts/tools.The Redis Operator provides a production-ready interface for managing Redis on Kubernetes with the following capabilities:
redis-exporter and detailed Grafana dashboards.StatefulSet (such as volumeClaimTemplates) are immutable. If you need to change these fields, set redisStandalone.recreateStatefulSetOnUpdateInvalid to true. This instructs the operator to delete the existing StatefulSet and recreate it to apply the changes. The default value is false.v0.21.0, when scaling in a Redis Cluster, the operator supports round-robin logic for transferring cluster shards. This ensures a more balanced distribution of data during scale-in operations.The Redis Operator supports two primary deployment models: Standalone and Cluster. When using Cluster mode, the operator currently focuses on Sharded Cluster setups.
In a replication model, data is mirrored from a leader node to follower nodes. Every node in the replication set contains a copy of the same data.
In a sharding model, data is split across multiple nodes based on keys. For example, Node A might hold keys 1 and 2, while Node B holds keys 3 and 4. This allows for horizontal scaling by distributing the data load across multiple machines.
The Redis Operator extends the Kubernetes API using Custom Resource Definitions (CRDs). Once the operator is installed, you can manage Redis instances by creating resources under the API group redis.redis.opstreelabs.in/v1beta2.
The following four object types are available for deployment and management:
The operator provides built-in monitoring using redis-exporter. It uses the Prometheus service discovery pattern by applying specific annotations to the pods.
To ensure Prometheus can scrape the metrics, the following annotations are used:
redis.opstreelabs.in: "true"prometheus.io/scrape: "true"prometheus.io/port: "9121"If you are using the Prometheus Operator, you can also deploy a ServiceMonitor for each Redis installation by configuring the appropriate values in your Helm values file.
annotations:
redis.opstreelabs.in: "true"
prometheus.io/scrape: "true"
prometheus.io/port: "9121"