Elasticsearch Integration
Cluster health, index metrics, JVM heap analysis, and query latency monitoring for Elasticsearch. Know when your cluster is under pressure before users notice slow searches.
How It Works
Configure the ES Exporter
Point the TigerOps Elasticsearch exporter at your cluster endpoint. It uses the _cluster/stats, _nodes/stats, and _cat APIs — no plugin installation required.
Set Up API Key Auth
Create a read-only Elasticsearch API key with monitor and read_cross_cluster privileges. TigerOps supports both API key and basic auth for secure, least-privilege access.
Configure Index Patterns
Specify which index patterns to monitor in detail. TigerOps tracks per-index shard health, document counts, store size, and query/indexing latency for your most important indices.
Receive Cluster Health Alerts
TigerOps fires alerts when cluster status transitions from green to yellow or red, when JVM heap exceeds safe thresholds, or when unassigned shards appear.
What You Get Out of the Box
Cluster Health & Status
Real-time cluster status (green/yellow/red), active shards, unassigned shards, relocating shards, and node count. TigerOps alerts immediately when health degrades.
Index & Shard Metrics
Per-index document count, store size, indexing rate, search rate, and merge throughput. Track index lifecycle policies and segment merging activity.
JVM Heap & GC Analysis
Per-node JVM heap used/max, old gen collection frequency, GC pause duration, and thread pool rejection rates. Alert before heap pressure causes circuit breaker trips.
Query Latency Percentiles
Search query latency P50, P95, and P99 per index and per node. TigerOps detects slow query patterns and correlates them with upstream application service spans.
Disk & I/O Monitoring
Per-node disk utilization, I/O wait, segment merge I/O, and translog flush rates. TigerOps warns when data nodes approach disk watermarks before Elasticsearch takes action.
Thread Pool & Circuit Breakers
Track thread pool queue depth and rejection counts for search, write, and management pools. Monitor circuit breaker trip counts to identify memory pressure events.
Elasticsearch Exporter Config
Configure the TigerOps Elasticsearch exporter to monitor your cluster.
# TigerOps Elasticsearch Exporter configuration
# Deploy via Helm: helm install es-exporter tigerops/elasticsearch-exporter -f values.yaml
es:
uri: https://your-es-cluster:9200
# API Key authentication (recommended)
apiKey: "${ES_API_KEY}"
# Or basic auth:
# username: tigerops_monitor
# password: "${ES_PASSWORD}"
ssl:
enabled: true
skipVerify: false # Set true only for internal CAs
# Index patterns to monitor in detail (glob supported)
indices:
- name: "orders-*"
- name: "users"
- name: "product-catalog"
# All other indices: aggregate metrics only
indicesAggregateOnly: false
# Collect shard-level metrics (expensive on large clusters)
shards: true
tigerops:
endpoint: https://ingest.atatus.net/api/v1/write
apiKey: "${TIGEROPS_API_KEY}"
scrapeInterval: 30s
# Alert thresholds (applied as alerting rules in TigerOps)
alerts:
clusterStatus:
yellow: warn
red: critical
jvmHeapUsedPercent:
warning: 75
critical: 90
unassignedShards:
critical: 1
diskWatermarkPercent:
warning: 80
critical: 85Common Questions
Does TigerOps support Elastic Cloud and OpenSearch?
Yes. TigerOps works with self-hosted Elasticsearch, Elastic Cloud (using Elastic API keys), and OpenSearch. OpenSearch uses the same REST API surface, so the exporter works without modification. AWS OpenSearch Service is also supported.
What privileges does the TigerOps Elasticsearch API key need?
The API key needs the monitor cluster privilege and read privilege on all indices you want to monitor. For cross-cluster monitoring: read_cross_cluster. No index write privileges are required.
Can TigerOps predict when Elasticsearch will run out of disk space?
Yes. TigerOps tracks disk usage growth rate per data node and forecasts when nodes will hit the high watermark. It fires a predictive alert before Elasticsearch begins moving shards, giving you time to act without a cluster disruption.
How does TigerOps handle large clusters with many indices?
You can configure index pattern filters in the exporter to collect detailed metrics for your most important indices and aggregate-only metrics for others. TigerOps handles clusters with thousands of indices efficiently.
Can I correlate slow Elasticsearch queries with my application traces?
Yes, if your application uses the TigerOps SDK or OTel instrumentation with Elasticsearch client spans. TigerOps links search operation spans to the per-index latency metrics, making it easy to identify which application queries are slow.
Keep Your Search Cluster Healthy
Cluster health, JVM analysis, and slow query detection. Predict problems before your cluster goes red.