Elasticsearch Metrics Dashboard for Cluster Monitoring

Use this dashboard to monitor your Elasticsearch clusters across cluster health, node resources, index operations, search and query performance, JVM and garbage collection, cache, circuit breakers, and thread pools.

Elasticsearch exposes cluster health, node stats, and index stats over its HTTP API, and the OpenTelemetry Collector's elasticsearch receiver scrapes them for this dashboard.

Dashboard Coverage

Use these panels to:

• Confirm cluster health: See cluster status, active/unassigned/relocating shards, pending tasks, and node count.
• Track node resources: Monitor CPU, disk, JVM heap and non-heap usage, open file descriptors, and HTTP connections per node.
• Measure index activity: Track document count, store size, segment count, operations completed and their time, and current merges.
• Watch search and query performance: Track query rate, average search latency, scroll and suggest rate, and query cache hit/miss rate.
• Inspect JVM and GC: Watch garbage collection time, thread count, classes loaded, OS memory, and load average.
• Review cache usage: Track cache memory usage, evictions, cache size, and hit/miss counts.
• Watch circuit breakers: Track estimated vs. limit memory per breaker and tripped count.
• Monitor thread pools: Track thread counts, queued tasks, and finished tasks.

Metrics Included

Cluster Health

• Cluster Health Status: Cluster health state (elasticsearch.cluster.health).
• Active Shards, Active Primary Shards, Unassigned Shards, Relocating Shards: Shard counts by state (elasticsearch.cluster.shards).
• Pending Tasks: Cluster-level changes not yet executed (elasticsearch.cluster.pending_tasks).
• Number of Nodes: Total nodes in the cluster (elasticsearch.cluster.nodes).
• In-flight Fetches: Unfinished fetches (elasticsearch.cluster.in_flight_fetch).

Node Metrics

• Node CPU Usage: OS-level CPU usage per node (elasticsearch.os.cpu.usage).
• Node Disk Usage: Available, free, and total disk space (elasticsearch.node.fs.disk.*).
• Node JVM Heap: Committed, max, and used heap (jvm.memory.heap.*).
• Node JVM Non-Heap Used: Non-heap and pool memory used (jvm.memory.nonheap.used, jvm.memory.pool.used).
• Node Open File Descriptors: Open files per node (elasticsearch.node.open_files).
• Node HTTP Connections: Open HTTP connections per node (elasticsearch.node.http.connections).

Index Metrics

• Index Document Count: Documents per index (elasticsearch.index.documents).
• Index Store Size: Shard store size on disk (elasticsearch.index.shards.size).
• Index Segments Count: Segment count per index (elasticsearch.index.segments.count).
• Index Operations Completed and Index Operations Time: Operation counts and time by type (elasticsearch.index.operations.completed, elasticsearch.index.operations.time).
• Index Current Merges: In-progress segment merges (elasticsearch.index.operations.merge.current).

Search & Query Performance

• Search Query Rate: Query operations per second (elasticsearch.index.operations.completed).
• Search Latency (Avg): Average query time derived from operation count and time.
• Scroll & Suggest Rate: Scroll and suggest operation rates.
• Query Cache Hit/Miss Rate: Query cache effectiveness (elasticsearch.node.cache.count).
• Query Cache Memory Usage: Query cache memory footprint (elasticsearch.node.cache.memory.usage).
• Cache Evictions: Query cache evictions (elasticsearch.node.cache.evictions).

JVM & Memory

• GC Collection Time: Time spent in garbage collection (jvm.gc.collections.elapsed).
• JVM Threads: Live thread count (jvm.threads.count).
• OS Memory (Free vs Used): Host memory usage (elasticsearch.os.memory).
• OS Load Average: 1, 5, and 15-minute load averages (elasticsearch.os.cpu.load_avg.*).
• JVM Classes Loaded: Loaded class count (jvm.classes.loaded).

Cache

• Node Cache Memory Usage and Node Cache Size: Cache memory and entry counts (elasticsearch.node.cache.memory.usage, elasticsearch.node.cache.size).
• Node Cache Evictions: Evictions per node (elasticsearch.node.cache.evictions).
• Cache Hit Count and Cache Miss Count: Get operation hits and misses (elasticsearch.node.operations.get.completed).

Circuit Breaker

• Breaker Memory: Estimated vs. limit memory per breaker (elasticsearch.breaker.memory.estimated, elasticsearch.breaker.memory.limit).
• Breaker Tripped Count: Times a breaker prevented an out-of-memory error (elasticsearch.breaker.tripped).

Thread Pool

• Thread Pool Threads: Threads per pool (elasticsearch.node.thread_pool.threads).
• Thread Pool Tasks Queued and Thread Pool Tasks Finished: Queued and completed tasks per pool (elasticsearch.node.thread_pool.tasks.queued, elasticsearch.node.thread_pool.tasks.finished).

Dashboard Variables

Use these filter variables:

• service.name: Filter by the OpenTelemetry service name.
• elasticsearch.cluster.name: Filter by cluster name.
• elasticsearch.node.name: Filter by node name.

Related Dashboards

• APM Metrics
• Docker Container