Skip to content

BACKEND_DEPLOYMENT_FIX.md

Latest commit

 

History

History
143 lines (110 loc) · 4.77 KB

BACKEND_DEPLOYMENT_FIX.md

File metadata and controls

143 lines (110 loc) · 4.77 KB
Raw

Backend Deployment Fix Summary

Issue Encountered

  1. 403 Error: Backend was using public Gemini API instead of Vertex AI
  2. Port Mismatch: Dockerfile set PORT=8000, Cloud Run expects PORT=8080
  3. Import Errors: Old incompatible files conflicting with new code

Fixes Applied

1. Vertex AI Configuration ✅

Problem: Backend defaulted to public Gemini API (generativelanguage.googleapis.com)
Root: use_vertex_ai: bool = False in settings.py

Changes:

Result: Backend now uses Vertex AI (aiplatform.googleapis.com) by default

2. Port Configuration ✅

Problem: Dockerfile set ENV PORT=8000, overriding Cloud Run's PORT=8080
Error: "Container failed to start and listen on the port defined by PORT=8080"

Changes:

  • Dockerfile: Removed ENV PORT=8000 line
  • Added comment explaining Cloud Run sets PORT=8080 automatically

Result: Backend now listens on Cloud Run's PORT (8080)

3. File Cleanup & Import Resolution ✅

Problem: Multiple versions of files causing import conflicts

Old incompatible files (renamed to .old.py):

  • backend/main.py → backend/main.old.py
  • backend/api/app.py → backend/api/app.old.py
  • backend/api/endpoints.py → backend/api/endpoints.old.py
  • backend/agents/agent_definitions.py → backend/agents/agent_definitions.old.py
  • backend/agents/agent_strategies.py → backend/agents/agent_strategies.old.py
  • backend/managers/chatbot_manager.py → backend/managers/chatbot_manager.old.py

New files activated (renamed from *_new.py → *.py):

  • backend/main_new.py → backend/main.py
  • backend/api/app_new.py → backend/api/app.py
  • backend/api/endpoints_new.py → backend/api/endpoints.py
  • backend/agents/agent_definitions_new.py → backend/agents/agent_definitions.py
  • backend/agents/agent_strategies_new.py → backend/agents/agent_strategies.py
  • backend/managers/chatbot_manager_new.py → backend/managers/chatbot_manager.py

Import updates in backend/api/app.py: -FROM:

from managers.chatbot_manager_new import get_chatbot_manager
from api.endpoints_new import router
from agents.agent_strategies_new import get_workflow_template
from agents.agent_definitions_new import get_agent_config

TO:

from managers.chatbot_manager import get_chatbot_manager
from api.endpoints import router
from agents.agent_strategies import get_workflow_template
from agents.agent_definitions import get_agent_config

Import updates in backend/main.py:

uvicorn.run("api.app:app", ...)  # Changed from "api.app_new:app"

Dockerfile CMD updated:

CMD ["python", "main.py"]  # Changed from "main_new.py"

Result: All imports resolved, no more conflicting file versions

Deployment Command

gcloud run deploy legalmind-backend \
  --source=backend \
  --project=legalmind-486106 \
  --region=us-central1 \
  --allow-unauthenticated \
  --set-env-vars="USE_VERTEX_AI=true" \
  --timeout=300

Expected Outcome

After deployment completes:

  1. ✅ Backend starts successfully on PORT=8080
  2. ✅ Health endpoint responds with HTTP 200
  3. ✅ Vertex AI configured correctly (no 403 errors)
  4. ✅ Contract analysis works without authentication errors

Test After Deployment

# Health check
curl https://legalmind-backend-677928716377.us-central1.run.app/health

# Expected response:
{
  "status": "healthy",
  "service": "LegalMind API",
  "timestamp": "..."
}

Files Modified

  1. Dockerfile - Removed PORT override, added USE_VERTEX_AI
  2. backend/config/settings.py - Changed defaults to use Vertex AI
  3. backend/.env.example - Documented USE_VERTEX_AI setting
  4. backend/main.py - Updated imports (renamed from main_new.py)
  5. backend/api/app.py - Updated imports (renamed from app_new.py)
  6. Multiple backend modules - Renamed from *_new.py to *.py

Summary

Root Cause: Configuration defaulted to public Gemini API + old incompatible files Solution: Enable Vertex AI by default + clean up file conflicts + fix port configuration
Status: Deployment in progress Deployment Time: ~3-5 minutes

Once deployment completes, the backend will:

  • Use Vertex AI exclusively (aiplatform.googleapis.com)
  • Listen on the correct port (8080)
  • Have no import conflicts
  • Work correctly with Cloud Run service accounts