feat: Dapr State Management for Caching & Session State #33

Summary

Use Dapr State Management API for caching frequently accessed data and session state, reducing database load and improving response times.

Current State

• All state reads go directly to PostgreSQL
• No caching layer for hot data
• Session/conversation state stored in DB
• Every request hits the database

Proposed Architecture

┌─────────────┐     ┌─────────────┐     ┌─────────────────┐
│ API Service │────►│ Dapr        │────►│ State Store     │
│             │     │ Sidecar     │     │ (Redis/Postgres)│
│             │     │ :3500       │     │                 │
└─────────────┘     └─────────────┘     └─────────────────┘

# Instead of:
task = await session.get(Task, task_id)

# Use Dapr State for hot data:
task = await dapr_client.get_state("statestore", f"task-{task_id}")

Use Cases

1. User Session Cache — Active user data, permissions
2. Project Membership — Frequently checked access control
3. Task Count/Stats — Dashboard metrics
4. Conversation State — Chat history for MCP
5. Rate Limiting — Request counters per user

Implementation Steps

• Add Dapr State component configuration (Redis-backed)
• Create dapr_state.py utility module
• Implement cache-aside pattern for tasks
• Add TTL-based expiration for cached items
• Cache user permissions on login
• Add cache invalidation on writes
• Implement distributed rate limiting

Acceptance Criteria

• Repeated reads for same task use cache (no DB hit)
• Cache invalidates correctly on task update
• Dashboard loads faster with cached stats
• Works with Dapr component swap (Redis → other store)
• Graceful fallback to DB if state store unavailable

Technical Notes

# Dapr State API example
from dapr.clients import DaprClient

async with DaprClient() as client:
    # Save state
    await client.save_state("statestore", key, value, state_metadata={"ttlInSeconds": "300"})
    
    # Get state
    state = await client.get_state("statestore", key)
    
    # Delete state
    await client.delete_state("statestore", key)

Labels

enhancement learning dapr performance phase-v2

📚 Learning Goal: Dapr State Management Building Block