Skip to main content

8.10 Root Cause Analysis

Root Cause Analysis (RCA) is an AI-powered investigative feature that, given an event, automatically retrieves relevant historical data, forms hypotheses about the cause, tests those hypotheses, and produces a structured analysis report — all without manual intervention.

This feature uses the Deep Thinking Model configured in Connection Management to perform multi-step reasoning over your time-series data, operational knowledge, and publicly available technical references.

How to Access​

RCA is accessed from the event detail page:

  1. Navigate to the Events section from the top navigation.
  2. Click any event to open its detail view.
  3. In the toolbar on the right side of the detail pane, click the Root Cause Analysis icon (the magnifying glass with a signal trace icon).

The Root Cause Analysis panel opens on the right side of the event detail page.

What Happens When You Start RCA​

The analysis runs as an automated multi-step workflow:

  1. Intent recognition — The system determines the analysis goal from the event context.
  2. Element confirmation — The associated element, device ID, occurrence time, and symptom are extracted from the event.
  3. Data retrieval — TDengine SQL is generated to fetch time-series data for the element from the surrounding time window (typically the last 10 days).
  4. Data exploration — Python analysis code is generated and executed to statistically explore the retrieved data, identifying outliers, trends, and correlations.
  5. Knowledge retrieval — A web search retrieves relevant technical documentation or known failure modes for this type of equipment.
  6. Hypothesis decomposition — The AI decomposes the problem into one or more sub-hypotheses about potential root causes.
  7. Hypothesis verification — Each hypothesis is tested against the actual data.
  8. Report generation — A structured Markdown report is generated summarizing all findings.

The panel streams the workflow progress in real time so you can follow each step as it executes.

The RCA Report​

The final report is a structured document that includes:

  • Overview — Event name, affected system scope, occurrence time, severity level
  • Timeline — A chronological table of key events and actions around the incident
  • Data Analysis — Statistical findings from the data exploration phase, including identified outliers or anomalies
  • Root Cause Hypotheses — The AI's ranked hypotheses about what caused the event, with supporting evidence from the data
  • Recommendations — Suggested corrective or preventive actions

Panel Controls​

ControlDescription
Refresh iconRe-run the root cause analysis for this event
Close (X)Close the RCA panel and return to the standard event detail view
note

Root Cause Analysis is a new feature introduced in TDengine IDMP 1.0.14. Additional entry points for RCA (such as from dashboards and the AI Chat interface) are planned for future releases.