Root Cause Analysis
The Open Distro for Elasticsearch Performance Analyzer plugin (PA) captures Elasticsearch and JVM activity, plus their lower-level resource usage (e.g. disk, network, CPU, and memory). Based on this instrumentation, Performance Analyzer computes and exposes diagnostic metrics so that administrators can measure and understand the bottlenecks in their Elasticsearch clusters.
The Root Cause Analysis framework (RCA) uses the information from PA to alert administrators about the root cause of performance and availability issues that their clusters might be experiencing.
In broad strokes, the framework helps you access data streams from Elasticsearch nodes running Performance Analyzer. You write snippets of Java to choose the streams that matter to you and evaluate the streams’ PA metrics against certain thresholds. As RCA runs, you can access the state of each analysis using the REST API.
To learn more about Root Cause Analysis, see its repository on GitHub.