fix(deploy): resolve exit code 3 — robust paths, --preload, log capture

- config.py: Use absolute path /home/site/wwwroot/data for SQLite on Azure,
  with /tmp fallback if mkdir fails (prevents PermissionError crash)
- startup.sh: Add 'exec > >(tee -a /home/LogFiles/startup.log) 2>&1' to
  capture ALL output to a persistent log file for debugging
- startup.sh: Add --preload to gunicorn so import errors are visible
  before worker forking (exit code 3 = worker boot failure)
- startup.sh: Test src.config import separately before src.api_server
  to isolate the failing layer
- deploy-azure.yml: Use absolute path 'bash /home/site/wwwroot/startup.sh'
  for startup command
- deploy-azure.yml: Increase health check retries to 5 with 20s delay
- deploy-azure.yml: Download and print Azure logs on health check failure

Author: Scarage1

.github/workflows/deploy-azure.yml

@@ -87,21 +87,32 @@ jobs:
           app-name: ${{ env.AZURE_WEBAPP_NAME }}
           package: backend-deploy.zip
           clean: true
-          startup-command: startup.sh
+          startup-command: "bash /home/site/wwwroot/startup.sh"
 
       - name: Test deployment
         run: |
           echo "Waiting for deployment to stabilize..."
-          sleep 30
-          for i in 1 2 3; do
+          sleep 60
+          for i in 1 2 3 4 5; do
             if curl -sf https://${{ env.AZURE_WEBAPP_NAME }}.azurewebsites.net/health; then
               echo "✅ Backend is healthy!"
               exit 0
             fi
-            echo "Attempt $i failed, retrying in 10s..."
-            sleep 10
+            echo "Attempt $i failed, retrying in 20s..."
+            sleep 20
           done
-          echo "❌ Health check failed after 3 attempts"
+          echo "❌ Health check failed after 5 attempts"
+          # Download logs for debugging
+          az webapp log download --resource-group ${{ env.AZURE_RESOURCE_GROUP }} \
+            --name ${{ env.AZURE_WEBAPP_NAME }} \
+            --log-file /tmp/azure-logs.zip 2>/dev/null || true
+          if [ -f /tmp/azure-logs.zip ]; then
+            unzip -o /tmp/azure-logs.zip -d /tmp/azure-logs 2>/dev/null || true
+            echo "=== Docker Logs ==="
+            cat /tmp/azure-logs/LogFiles/*docker*.log 2>/dev/null | tail -80 || true
+            echo "=== Startup Log ==="
+            cat /tmp/azure-logs/LogFiles/startup.log 2>/dev/null || true
+          fi
           exit 1
 
       - name: Logout from Azure

src/config.py

@@ -81,9 +81,18 @@ class Settings(BaseSettings):
     def _default_database_url(cls, v: str) -> str:
         if v:
             return v
-        # Default to SQLite in ./data directory
-        db_dir = Path("data")
-        db_dir.mkdir(parents=True, exist_ok=True)
+        # Default to SQLite — use /home/site/wwwroot/data on Azure,
+        # otherwise ./data relative to project root
+        if os.path.isdir("/home/site/wwwroot"):
+            db_dir = Path("/home/site/wwwroot/data")
+        else:
+            db_dir = Path("data")
+        try:
+            db_dir.mkdir(parents=True, exist_ok=True)
+        except OSError as e:
+            logger.warning("Cannot create DB directory %s: %s — using /tmp", db_dir, e)
+            db_dir = Path("/tmp/apiwatch-data")
+            db_dir.mkdir(parents=True, exist_ok=True)
         return f"sqlite+aiosqlite:///{db_dir.resolve()}/apiwatch.db"
 
     @field_validator("jwt_secret_key", mode="before")

startup.sh

@@ -2,65 +2,79 @@
 # Azure App Service startup script
 # Azure sets PORT env var automatically (default 8000)
 
+# Redirect ALL output to both stdout and a log file for debugging
+exec > >(tee -a /home/LogFiles/startup.log) 2>&1
+
 echo "=== API-Watch Startup ==="
 echo "PWD: $(pwd)"
 echo "Date: $(date)"
 
 # Ensure we're in the deployment directory
-cd /home/site/wwwroot
+cd /home/site/wwwroot || { echo "FATAL: cannot cd to /home/site/wwwroot"; exit 1; }
 echo "Working dir: $(pwd)"
-echo "Contents: $(ls -la)"
+echo "Contents:"
+ls -la
 
 # Activate Oryx-created virtual environment (created during zip deployment)
 if [ -d "antenv" ]; then
   echo "Activating Oryx virtual environment (antenv)..."
   source antenv/bin/activate
   echo "Python: $(which python)"
-  echo "Pip packages: $(pip list --format=columns 2>/dev/null | head -5)"
+  echo "Python version: $(python --version 2>&1)"
 else
   echo "WARNING: antenv directory not found!"
-  echo "Checking for packages directory..."
-fi
-
-# Add bundled packages to Python path (fallback if no antenv)
-if [ -d "packages" ]; then
-  export PYTHONPATH="/home/site/wwwroot/packages:$PYTHONPATH"
-  echo "Using pre-bundled Python packages"
+  echo "Listing site directory:"
+  ls -la /home/site/wwwroot/
 fi
 
-# Fallback: install if neither antenv nor bundled packages have gunicorn
+# Fallback: install if antenv doesn't have gunicorn
 if ! python -c "import gunicorn" 2>/dev/null; then
   echo "gunicorn not found, installing dependencies..."
-  pip install --no-cache-dir -r requirements.txt 2>&1 | tail -10 || echo "⚠️  pip install failed"
+  pip install --no-cache-dir -r requirements.txt 2>&1 | tail -10 || echo "pip install failed"
 fi
 
 # Verify critical imports work
-echo "Testing critical imports..."
+echo "=== Testing critical imports ==="
 python -c "
 import sys
 print(f'Python: {sys.executable}')
+print(f'sys.path: {sys.path[:5]}')
 try:
-    import fastapi; print(f'fastapi: {fastapi.__version__}')
-except Exception as e: print(f'fastapi FAILED: {e}')
+    import fastapi; print(f'  fastapi: {fastapi.__version__}')
+except Exception as e: print(f'  fastapi FAILED: {e}')
 try:
-    import gunicorn; print(f'gunicorn: OK')
-except Exception as e: print(f'gunicorn FAILED: {e}')
+    import gunicorn; print(f'  gunicorn: OK')
+except Exception as e: print(f'  gunicorn FAILED: {e}')
 try:
-    from src.api_server import app; print('src.api_server: OK')
-except Exception as e: print(f'src.api_server FAILED: {e}')
-" 2>&1 || echo "Import test failed"
+    import uvicorn; print(f'  uvicorn: OK')
+except Exception as e: print(f'  uvicorn FAILED: {e}')
+try:
+    from src.config import get_settings
+    s = get_settings()
+    print(f'  src.config: OK (db={s.database_url[:30]}...)')
+except Exception as e: print(f'  src.config FAILED: {e}')
+try:
+    from src.api_server import app; print(f'  src.api_server: OK')
+except Exception as e:
+    print(f'  src.api_server FAILED: {e}')
+    import traceback; traceback.print_exc()
+" 2>&1
+
+echo "=== Import test complete ==="
 
 # Run database migrations (safe to run repeatedly — no-ops if up to date)
 if [ -d "alembic" ]; then
   echo "Running database migrations..."
-  python -m alembic upgrade head 2>&1 || echo "⚠️  Migrations skipped (non-fatal)"
+  python -m alembic upgrade head 2>&1 || echo "Migrations skipped (non-fatal)"
 fi
 
 # Start with gunicorn + uvicorn workers for production performance
+# --preload: load app before forking so import errors are visible in main process
 export PORT=${PORT:-8000}
-echo "Starting API-Watch on port $PORT..."
+echo "Starting gunicorn on port $PORT with --preload..."
 
 exec gunicorn src.api_server:app \
+  --preload \
   --worker-class uvicorn.workers.UvicornWorker \
   --bind "0.0.0.0:$PORT" \
   --workers 1 \