Skip to content

demo.md

Latest commit

 

History

History
72 lines (59 loc) · 3.06 KB

demo.md

File metadata and controls

72 lines (59 loc) · 3.06 KB
Raw

RCA Prototype Demo

This prototype demonstrates an AI-enhanced Root Cause Analysis (RCA) system for Managed Service Providers (MSPs), integrating coroot with Keep alerts, AWS Bedrock AgentCore, and Strands Agents.

Features Highlighted

  1. Keep Integration: Alerts are sent to Keep for centralized alert management.
  2. AWS Bedrock AgentCore: Orchestrates alert processing with AI agents.
  3. Strands Agents: Provides autonomous workflows for drift detection and summarization.
  4. AI-Enhanced RCA: RCA views include agent-driven insights.
  5. MSP Frontend: Custom UI for client management and multi-tenant support.
  6. Multi-Tenant Support: Projects are isolated by client_id.

Setup Instructions

Prerequisites

  • Go 1.23+
  • Node.js for frontend
  • Docker for deployment (optional)
  • AWS credentials for Bedrock (optional, mocked if not configured)
  • Keep server running (optional, mocked if not configured)

Backend Setup

  1. Clone the repository: git clone https://github.com/coroot/coroot.git
  2. Navigate to coroot: cd coroot
  3. Install dependencies: go mod tidy
  4. Build: go build
  5. (Optional) Configure config.yaml with Keep URL, AWS settings; if not configured, mocking is used.

Frontend Setup

  1. Navigate to front: cd front
  2. Install dependencies: npm install
  3. Build: npm run build

Deployment

Local

  1. Build Docker image: docker build -t rca-prototype .
  2. Run: docker run -p 8080:8080 rca-prototype

Production

  • Frontend (Netlify): Deploy coroot/front/dist directory to Netlify. Set environment variable VUE_APP_API_BASE_URL to Render backend URL.
  • Backend (Render): Deploy Go app from coroot/ directory to Render. Use build command go build -o rca-backend. Set start command ./rca-backend. Configure environment variables like KEEP_URL, BEDROCK_AGENT_ID, etc. (leave empty for mock).

Demo Steps

  1. Start the Application: Run the backend server.
  2. Access Frontend: Open http://localhost:8080
  3. Create Project: Add a project with client_id for multi-tenancy.
  4. Configure Integrations: Set up Prometheus, Keep, AWS Bedrock.
  5. Simulate Incident: Trigger an alert in the system.
  6. View RCA: Navigate to RCA view for an application.
    • See AI Agent Insights section with Strands summarization.
  7. Check Logs: Verify mock alerts logged (if Keep not configured).
  8. Client Management: Access Client Management view for MSP features.

Key Files Modified

  • keep/keep.go: Keep client for alerts.
  • bedrock/bedrock.go: AWS Bedrock integration.
  • strands/strands.go: Strands agents for workflows.
  • api/rca.go: Extended with agent insights.
  • views/RCA.vue: Added AI insights display.
  • views/ClientManagement.vue: MSP client management.
  • db/project.go: Added client_id for multi-tenancy.

Validation Checklist

  • Backend compiles without errors.
  • Frontend builds successfully.
  • Docker image builds.
  • Application starts and serves UI.
  • RCA shows agent insights.
  • Mock alerts logged (if Keep not configured).
  • Multi-tenant isolation works.