check_modal_config.py•4.37 kB
#!/usr/bin/env python3
"""
Modal Configuration Safety Checker
Verifies modal_app.py is configured safely (CPU-only, no accidental GPU usage)
"""
import re
import sys
def check_modal_config():
"""Check modal_app.py for safe configuration"""
print("=" * 60)
print("Modal Configuration Safety Check")
print("=" * 60)
print()
try:
with open("modal_app.py", "r") as f:
content = f.read()
except FileNotFoundError:
print("❌ modal_app.py not found!")
return False
checks_passed = True
# Split content by function definitions
functions = re.split(r'@app\.function', content)
print("📋 Found Functions:")
print()
# Check 1: GPU functions and their limits
gpu_functions = []
for i, func in enumerate(functions):
if 'gpu=' in func:
# Extract function name
func_name_match = re.search(r'def\s+(\w+)', func)
func_name = func_name_match.group(1) if func_name_match else f"function_{i}"
gpu_functions.append(func_name)
# Check GPU concurrency limit
gpu_concurrency = re.search(r'concurrency_limit\s*=\s*([0-9]+)', func)
gpu_type = re.search(r'gpu\s*=\s*["\']([^"\']+)["\']', func)
scaledown = re.search(r'scaledown_window\s*=\s*([0-9]+)', func)
print(f"🎮 GPU Function: {func_name}")
print(f" GPU Type: {gpu_type.group(1) if gpu_type else 'unspecified'}")
if gpu_concurrency:
limit = int(gpu_concurrency.group(1))
if limit <= 2:
print(f" ✅ Concurrency: {limit} GPU(s) max (SAFE)")
else:
print(f" ⚠️ Concurrency: {limit} GPUs (HIGH!)")
checks_passed = False
else:
print(" ❌ No concurrency limit! (DANGEROUS)")
checks_passed = False
if scaledown:
window = int(scaledown.group(1))
print(f" ✅ Scaledown: {window}s idle → shutdown")
else:
print(" ⚠️ No scaledown window")
print()
if not gpu_functions:
print("✓ No GPU functions found (CPU-only deployment)")
# Check 2: Main serve function is CPU-only
serve_func = [f for f in functions if 'def serve' in f]
if serve_func:
if 'gpu=' in serve_func[0]:
print("❌ CRITICAL: Main serve() function has GPU! (Very expensive!)")
checks_passed = False
else:
print("✅ Main serve() function: CPU-only")
else:
print("⚠️ No serve() function found")
print()
print("=" * 60)
print("💰 Cost Estimate:")
print("=" * 60)
# Calculate max cost
if gpu_functions:
gpu_concurrency = re.findall(r'concurrency_limit\s*=\s*([0-9]+)', content)
if gpu_concurrency:
max_gpus = sum(int(x) for x in gpu_concurrency if int(x) <= 10) # Only count reasonable limits
print(f" Max GPUs at once: {max_gpus}")
print(f" Max GPU cost: ${max_gpus * 2.16:.2f}/hour (worst case)")
print(f" GPU idle cost: $0.036/minute after training")
print()
print(f" 💡 GPU shuts down after 60s idle → $0 when not used")
print(f" CPU service: ~$0.36/hour (always on)")
print(f" Free tier: 30 CPU-hours/month")
print()
print("=" * 60)
if checks_passed:
print("✅ Configuration looks SAFE!")
print()
print(" ✓ Main service: CPU-only")
if gpu_functions:
print(f" ✓ GPU functions: Limited to {gpu_concurrency[0] if gpu_concurrency else 'N/A'} concurrent")
print(" ✓ Auto-shutdown enabled")
print(" ✓ Cost-controlled")
print()
print("🚀 Safe to deploy!")
else:
print("❌ CONFIGURATION HAS ISSUES!")
print(" Please fix the issues above before deploying.")
print()
print("⛔ DO NOT DEPLOY until fixed!")
print("=" * 60)
print()
return checks_passed
if __name__ == "__main__":
safe = check_modal_config()
sys.exit(0 if safe else 1)