AIRA-59: Complete Analysis Workflow Integration

[learn-design-develop]

👤 User Story

As the orchestrator, I need the end-to-end analysis workflow that integrates all core components with error handling, data collection, and notifications.

🎯 Rationale

Ensures cohesive workflow from webhook reception to notification delivery.

✅ Acceptance Criteria

Implement run_analysis() in aira/core/workflow_integration.py with stages:

• Webhook reception & validation
• Deduplication (Redis)
• State transition (new → processing)
• Multi-source data collection (parallel, timeout DATA_COLLECTION_TIMEOUT)
  • Implement multi-source data collection in parallel using Python’s asyncio.gather (or equivalent), ensuring that all connector fetches (PagerDuty, GitHub, Datadog, etc.) are awaited concurrently within the run_analysis() method.
• On individual connector timeout or error, log the failure, omit that data source, and continue workflow execution.
• LLM analysis (OpenAIProvider.generate_hypothesis())
• Persist AnalysisResult; update state → analyzed
• Notification delivery to Slack
• Final state update → closed; metrics (workflow_duration_seconds, workflow_executions_total)
  • Stage-specific retries (max 3)
  • Circuit breaker integration
  • Dead-letter logging
  • Overall timeout (WORKFLOW_TIMEOUT)
  • Audit logging with correlation IDs
  • Integration tests covering success & failure

📋 Metadata

• Status: MVP

• Category: Core Workflow

• Week: Week 5

• Complexity: High

• Critical Path: Yes

• Dependencies: AIRA-43, AIRA-53, AIRA-28

---

Original Ticket: #59
Phase 1 MVP Tracking Issue