Shard Health Microscope
The cat shards API provides detailed visibility into shard allocation, distribution patterns, and health status across all indices. It's your window into the data layer's operational state.
Shard distribution is the foundation of Elasticsearch's scalability and resilience. The cat shards check examines every shard in your cluster, analyzing allocation patterns, identifying problematic shards, and ensuring optimal data distribution for performance and availability.
Cat Shards API Deep Dive
GET /_cat/shards?v&h=index,shard,prirep,state,docs,store,ip,node&s=index🔍 Shard Information
- • State: STARTED, INITIALIZING, RELOCATING, UNASSIGNED
- • Type: Primary (p) or Replica (r)
- • Size: Storage utilization per shard
- • Documents: Document count per shard
📊 Distribution Analysis
- • Node Allocation: Which node hosts each shard
- • Balance: Even distribution across nodes
- • Hotspots: Overloaded nodes identification
- • Availability: Replica placement validation
ElasticDoctor Shard Analysis
🔍 How ElasticDoctor Analyzes Shard Distribution
Distribution Balance Analysis
ElasticDoctor evaluates shard distribution across nodes to identify imbalances and hotspots that could impact performance and availability.
Unassigned Shard Detection
Automatically identifies unassigned shards and categorizes them by severity, with immediate alerts for unassigned primary shards that affect data availability.
Shard Size Optimization
Monitors shard sizes to identify oversized shards that may impact performance and undersized shards that waste resources.
Allocation Health Scoring
Provides comprehensive allocation health scoring based on distribution balance, shard states, and replica placement patterns.
Shard Management Best Practices
✅ Optimal Distribution
- • Aim for even shard distribution across nodes
- • Keep shard sizes between 10-50GB for optimal performance
- • Ensure primary and replica shards on different nodes
- • Monitor and maintain balanced storage utilization
- • Use allocation awareness for rack/zone distribution
💡 Monitoring Strategy
- • Regular shard health checks
- • Alert on unassigned shard conditions
- • Track shard size growth over time
- • Monitor rebalancing operations
- • Validate replica placement
❌ Distribution Anti-Patterns
- • Too many small shards (<1GB)
- • Oversized shards (>50GB)
- • All shards on single node
- • Primary and replica on same node
- • Ignoring unassigned shards
⚠️ Performance Impact
- • Unbalanced load affects query performance
- • Large shards slow search operations
- • Too many shards increase overhead
- • Relocation impacts cluster performance
- • Unassigned shards cause data unavailability
Shard Health Mastery
Distribution Excellence
- • Balanced Allocation: Even distribution across nodes
- • Optimal Sizing: Right-sized shards for performance
- • High Availability: Proper replica placement
- • Proactive Monitoring: Early issue detection
Operational Excellence
- • Monitor shard states continuously
- • Address unassigned shards immediately
- • Maintain balanced distribution
- • Plan for capacity and growth