Skip to main content

mr-cassop Architecture

mr-cassop is a comprehensive Kubernetes operator that deploys and manages Cassandra clusters along with essential supporting components for monitoring, backup, and maintenance.

Core Components​

A CassandraCluster managed by mr-cassop consists of the following integrated components:

🎯 Operator Controller​

  • Purpose: Main reconciliation engine that manages cluster state
  • Responsibilities:
    • Watches CassandraCluster resources
    • Creates and manages StatefulSets, Services, and ConfigMaps
    • Handles scaling operations and rolling updates
    • Coordinates with other components
  • Deployment: Runs as a Deployment in the mr-cassop-system namespace

🗄️ Cassandra Nodes​

  • Purpose: The core Apache Cassandra database instances
  • Deployment: Managed by StatefulSets (one per datacenter)
  • Features:
    • Persistent storage for data
    • Automatic cluster formation
    • JMX monitoring enabled
    • Authentication and authorization configured

🔍 Prober​

  • Purpose: Health monitoring and cross-region communication
  • Responsibilities:
    • Readiness and liveness checks
    • Metrics collection and exposition
    • Inter-datacenter health monitoring
    • Node status aggregation
  • Deployment: Sidecar container in each Cassandra pod
  • Documentation: Prober Details

📊 Jolokia​

  • Purpose: JMX-to-HTTP bridge for metrics collection
  • Responsibilities:
    • Exposes Cassandra JMX metrics over HTTP
    • Provides secure access to management operations
    • Enables monitoring integrations
  • Deployment: Sidecar container in each Cassandra pod
  • Website: jolokia.org

🔧 Reaper​

  • Purpose: Automated repair management and scheduling
  • Responsibilities:
    • Cluster repair coordination
    • Repair schedule management
    • Repair progress monitoring
    • Performance optimization
  • Deployment: Separate Deployment per cluster
  • Website: cassandra-reaper.io
  • Documentation: Reaper Configuration

💾 Icarus​

  • Purpose: Backup and restore service
  • Responsibilities:
    • Automated backup creation
    • Point-in-time recovery
    • Cross-region backup replication
    • Backup lifecycle management
  • Deployment: Sidecar container in each Cassandra pod

System Architecture​

┌─────────────────────────────────────────────────────────────────┐
│                        Kubernetes Cluster                      │
├─────────────────────────────────────────────────────────────────┤
│  mr-cassop-system namespace                                     │
│  ┌─────────────────┐                                           │
│  │   mr-cassop     │                                           │
│  │   (Operator)    │ ◄────── Watches CassandraCluster CRDs    │
│  └─────────────────┘                                           │
├─────────────────────────────────────────────────────────────────┤
│  Application namespace (e.g., cassop)                          │
│                                                                 │
│  ┌─────────────────┐    ┌─────────────────┐                   │
│  │     Reaper      │    │   Monitoring    │                   │
│  │   (Repairs)     │    │   Dashboard     │                   │
│  └─────────────────┘    └─────────────────┘                   │
│                                                                 │
│  ┌─────────────────────────────────────────────────────────┐   │
│  │              Cassandra StatefulSet (dc1)               │   │
│  │                                                         │   │
│  │  ┌─────────┐  ┌─────────┐  ┌─────────┐                 │   │
│  │  │  Pod-0  │  │  Pod-1  │  │  Pod-2  │                 │   │
│  │  │         │  │         │  │         │                 │   │
│  │  │ ┌─────┐ │  │ ┌─────┐ │  │ ┌─────┐ │                 │   │
│  │  │ │ C*  │ │  │ │ C*  │ │  │ │ C*  │ │                 │   │
│  │  │ └─────┘ │  │ └─────┘ │  │ └─────┘ │                 │   │
│  │  │ ┌─────┐ │  │ ┌─────┐ │  │ ┌─────┐ │                 │   │
│  │  │ │Prober│ │  │ │Prober│ │  │ │Prober│ │                 │   │
│  │  │ └─────┘ │  │ └─────┘ │  │ └─────┘ │                 │   │
│  │  │ ┌─────┐ │  │ ┌─────┐ │  │ ┌─────┐ │                 │   │
│  │  │ │Jolokia│ │  │ │Jolokia│ │  │ │Jolokia│ │                 │   │
│  │  │ └─────┘ │  │ └─────┘ │  │ └─────┘ │                 │   │
│  │  │ ┌─────┐ │  │ ┌─────┐ │  │ ┌─────┐ │                 │   │
│  │  │ │Icarus│ │  │ │Icarus│ │  │ │Icarus│ │                 │   │
│  │  │ └─────┘ │  │ └─────┘ │  │ └─────┘ │                 │   │
│  │  └─────────┘  └─────────┘  └─────────┘                 │   │
│  └─────────────────────────────────────────────────────────┘   │
└─────────────────────────────────────────────────────────────────┘

Component Interactions​

  1. Operator Controller continuously watches for CassandraCluster resources
  2. StatefulSets are created/updated based on cluster specifications
  3. Prober containers monitor Cassandra health and report status
  4. Jolokia exposes JMX metrics for monitoring and management
  5. Reaper automatically schedules and executes repairs
  6. Icarus handles backup operations based on configured schedules
  7. Services provide stable networking for client connections

Multi-Datacenter Support​

For multi-region deployments, mr-cassop creates:

  • One StatefulSet per datacenter
  • Cross-datacenter networking configuration
  • Automatic seed node management
  • Region-aware backup strategies

See Multi-Region Configuration for details.

Monitoring Integration​

The architecture supports comprehensive monitoring through:

  • Prometheus metrics from Jolokia and Prober
  • Grafana dashboards for visualization
  • Kubernetes native health checks
  • Custom alerts and notifications

See Prober Documentation for monitoring and health-check details.

Last updated on