gitarsenal-cli 1.6.5 → 1.6.8

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "gitarsenal-cli",
-  "version": "1.6.5",
+  "version": "1.6.8",
   "description": "CLI tool for creating Modal sandboxes with GitHub repositories",
   "main": "index.js",
   "bin": {

package/python/cuda_image_options.py

@@ -0,0 +1,124 @@
+#!/usr/bin/env python3
+"""
+Alternative CUDA image options for GitArsenal CLI
+These images are more stable and less likely to cause segmentation faults
+"""
+
+import modal
+
+def get_stable_cuda_image():
+    """
+    Get a stable CUDA image that's less likely to cause segmentation faults
+    """
+    return modal.Image.from_registry("nvidia/cuda:11.8.0-runtime-ubuntu22.04", add_python="3.11")
+
+def get_lightweight_cuda_image():
+    """
+    Get a lightweight CUDA image for basic GPU operations
+    """
+    return modal.Image.from_registry("nvidia/cuda:11.8.0-base-ubuntu22.04", add_python="3.11")
+
+def get_latest_stable_cuda_image():
+    """
+    Get the latest stable CUDA image (12.1 instead of 12.4)
+    """
+    return modal.Image.from_registry("nvidia/cuda:12.1.0-runtime-ubuntu22.04", add_python="3.11")
+
+def get_minimal_cuda_image():
+    """
+    Get a minimal CUDA image with just the essentials
+    """
+    return modal.Image.from_registry("nvidia/cuda:11.8.0-minimal-ubuntu22.04", add_python="3.11")
+
+def get_custom_cuda_image():
+    """
+    Create a custom CUDA image with specific optimizations
+    """
+    return (
+        modal.Image.from_registry("nvidia/cuda:11.8.0-runtime-ubuntu22.04", add_python="3.11")
+        .apt_install(
+            "openssh-server", "sudo", "curl", "wget", "vim", "htop", "git", 
+            "python3", "python3-pip", "build-essential", "tmux", "screen", "nano",
+            "gpg", "ca-certificates", "software-properties-common"
+        )
+        .pip_install("uv", "modal", "requests", "openai")
+        .run_commands(
+            # SSH setup
+            "mkdir -p /var/run/sshd",
+            "mkdir -p /root/.ssh",
+            "chmod 700 /root/.ssh",
+            "sed -i 's/#PermitRootLogin prohibit-password/PermitRootLogin yes/' /etc/ssh/sshd_config",
+            "sed -i 's/#PasswordAuthentication yes/PasswordAuthentication yes/' /etc/ssh/sshd_config",
+            "sed -i 's/#PubkeyAuthentication yes/PubkeyAuthentication yes/' /etc/ssh/sshd_config",
+            "echo 'ClientAliveInterval 60' >> /etc/ssh/sshd_config",
+            "echo 'ClientAliveCountMax 3' >> /etc/ssh/sshd_config",
+            "ssh-keygen -A",
+            
+            # GPU compatibility
+            "echo 'export CUDA_VISIBLE_DEVICES=0' >> /root/.bashrc",
+            "echo 'export NVIDIA_VISIBLE_DEVICES=all' >> /root/.bashrc",
+            "echo 'export NVIDIA_DRIVER_CAPABILITIES=compute,utility' >> /root/.bashrc",
+            
+            # Bash prompt
+            "echo 'export PS1=\"\\[\\e[1;32m\\]modal:\\[\\e[1;34m\\]\\w\\[\\e[0m\\]$ \"' >> /root/.bashrc",
+        )
+    )
+
+# Image selection based on use case
+CUDA_IMAGE_OPTIONS = {
+    "stable": get_stable_cuda_image,
+    "lightweight": get_lightweight_cuda_image,
+    "latest": get_latest_stable_cuda_image,
+    "minimal": get_minimal_cuda_image,
+    "custom": get_custom_cuda_image,
+    "default": lambda: modal.Image.debian_slim()  # No CUDA, most stable
+}
+
+def get_cuda_image(option="default"):
+    """
+    Get a CUDA image based on the specified option
+    
+    Args:
+        option (str): One of "stable", "lightweight", "latest", "minimal", "custom", "default"
+    
+    Returns:
+        modal.Image: The selected CUDA image
+    """
+    if option not in CUDA_IMAGE_OPTIONS:
+        print(f"⚠️ Unknown CUDA image option: {option}. Using default.")
+        option = "default"
+    
+    return CUDA_IMAGE_OPTIONS[option]()
+
+def test_cuda_image_stability(image_func, name):
+    """
+    Test the stability of a CUDA image
+    
+    Args:
+        image_func: Function that returns a modal.Image
+        name (str): Name of the image for logging
+    
+    Returns:
+        bool: True if image builds successfully
+    """
+    try:
+        print(f"🧪 Testing {name} CUDA image...")
+        image = image_func()
+        print(f"✅ {name} image created successfully")
+        return True
+    except Exception as e:
+        print(f"❌ {name} image failed: {e}")
+        return False
+
+if __name__ == "__main__":
+    print("🧪 Testing CUDA image stability...")
+    
+    for name, image_func in CUDA_IMAGE_OPTIONS.items():
+        test_cuda_image_stability(image_func, name)
+    
+    print("\n📋 CUDA Image Recommendations:")
+    print("• For maximum stability: Use 'default' (no CUDA)")
+    print("• For basic GPU operations: Use 'stable' (CUDA 11.8 runtime)")
+    print("• For minimal GPU support: Use 'minimal' (CUDA 11.8 minimal)")
+    print("• For latest features: Use 'latest' (CUDA 12.1 runtime)")
+    print("• For custom setup: Use 'custom' (CUDA 11.8 with SSH)") 

package/python/fix_modal_token.py

@@ -51,14 +51,14 @@ try:
     print(f"✅ Using tokens from proxy server or defaults")
 except (ImportError, ValueError) as e:
     # If the module is not available or tokens are invalid, use hardcoded tokens
-    # print(f"⚠️ Using default tokens")
+    print("")
 
 # print("🔧 Fixing Modal token (basic implementation)...")
 
 # Set environment variables
 # os.environ["MODAL_TOKEN_ID"] = TOKEN_ID
-os.environ["MODAL_TOKEN_SECRET"] = TOKEN_SECRET
-print(f"✅ Set MODAL_TOKEN_ID and MODAL_TOKEN_SECRET environment variables")
+# os.environ["MODAL_TOKEN_SECRET"] = TOKEN_SECRET
+# print(f"✅ Set MODAL_TOKEN_ID and MODAL_TOKEN_SECRET environment variables")
 
 # Create token file
 modal_dir = Path.home() / ".modal"

package/python/fix_modal_token_advanced.py

@@ -28,6 +28,7 @@ try:
 except ImportError:
     # If the module is not available, use hardcoded tokens
     # print(f"⚠️ Using default tokens")
+    print("")
 
 # print("🔧 Advanced Modal Token Fixer")
 
@@ -118,6 +119,7 @@ try:
                 # print(f"✅ Set tokens via _auth_config")
         except Exception as e:
             # print(f"❌ Error setting tokens via _auth_config: {e}")
+            print("")
             
         try:
             # Approach 4.2: Set token via set_token() if it exists
@@ -126,7 +128,7 @@ try:
                 # print(f"✅ Set tokens via set_token()")
         except Exception as e:
             # print(f"❌ Error setting tokens via set_token(): {e}")
-            
+            print("")
         try:
             # Approach 4.3: Set token via Config
             if hasattr(modal.config, 'Config'):
@@ -135,6 +137,7 @@ try:
                 # print(f"✅ Set tokens via Config")
         except Exception as e:
             # print(f"❌ Error setting tokens via Config: {e}")
+            print("")
             
         # Approach 4.4: Inspect modal.config and try to find token-related attributes
         # print("\n🔍 Inspecting modal.config for token-related attributes...")
@@ -151,8 +154,10 @@ try:
                         setattr(attr, "token_secret", TOKEN_SECRET)
                 except Exception as e:
                     # print(f"  - Error setting tokens in {name}: {e}")
+                    print("")
     except Exception as e:
         # print(f"❌ Error setting tokens in Modal config: {e}")
+        print("")
 except Exception as e:
     print(f"❌ Error importing Modal: {e}")
 
@@ -192,6 +197,7 @@ try:
                                 setattr(module, func_name, get_token_id)
                         except Exception as e:
                             # print(f"  - Error patching {name}.{func_name}: {e}")
+                            print("")
                     elif "token_secret" in func_name.lower() or "token" in func_name.lower():
                         try:
                             original_func = getattr(module, func_name)
@@ -200,11 +206,12 @@ try:
                                 setattr(module, func_name, get_token_secret)
                         except Exception as e:
                             # print(f"  - Error patching {name}.{func_name}: {e}")
+                            print("")
     
     # print(f"✅ Monkey-patching completed")
 except Exception as e:
     # print(f"❌ Error during monkey-patching: {e}")
-
+    print("")
 # Approach 6: Test Modal authentication
 # print("\n📋 Approach 6: Testing Modal authentication")
 try:
@@ -227,5 +234,6 @@ try:
         
 except Exception as e:
     # print(f"❌ Error testing Modal authentication: {e}")
+    print("")
 
 # print("\n✅ Done fixing Modal token. Please try your command again.") 

package/python/test_modalSandboxScript.py

@@ -968,218 +968,6 @@ def generate_random_password(length=16):
     password = ''.join(secrets.choice(alphabet) for i in range(length))
     return password
 
-# First, add the standalone ssh_container function at the module level, before the create_modal_ssh_container function
-
-# Define a module-level ssh container function
-ssh_app = modal.App("ssh-container-app")
-
-@ssh_app.function(
-    image=modal.Image.debian_slim()
-        .apt_install(
-            "openssh-server", "sudo", "curl", "wget", "vim", "htop", "git", 
-            "python3", "python3-pip", "build-essential", "tmux", "screen", "nano",
-            "gpg", "ca-certificates", "software-properties-common"
-        )
-        .pip_install("uv", "modal", "requests", "openai")  # Fast Python package installer and Modal
-        .run_commands(
-            # Create SSH directory
-            "mkdir -p /var/run/sshd",
-            "mkdir -p /root/.ssh",
-            "chmod 700 /root/.ssh",
-            
-            # Configure SSH server
-            "sed -i 's/#PermitRootLogin prohibit-password/PermitRootLogin yes/' /etc/ssh/sshd_config",
-            "sed -i 's/#PasswordAuthentication yes/PasswordAuthentication yes/' /etc/ssh/sshd_config",
-            "sed -i 's/#PubkeyAuthentication yes/PubkeyAuthentication yes/' /etc/ssh/sshd_config",
-            
-            # SSH keep-alive settings
-            "echo 'ClientAliveInterval 60' >> /etc/ssh/sshd_config",
-            "echo 'ClientAliveCountMax 3' >> /etc/ssh/sshd_config",
-            
-            # Generate SSH host keys
-            "ssh-keygen -A",
-            
-            # Install Modal CLI
-            "pip install modal",
-            
-            # Set up a nice bash prompt
-            "echo 'export PS1=\"\\[\\e[1;32m\\]modal:\\[\\e[1;34m\\]\\w\\[\\e[0m\\]$ \"' >> /root/.bashrc",
-        ),
-    timeout=3600,  # Default 1 hour timeout
-    gpu="a10g",    # Default GPU - this will be overridden when called
-    cpu=2,
-    memory=8192,
-    serialized=True,
-)
-def ssh_container_function(ssh_password, repo_url=None, repo_name=None, setup_commands=None, openai_api_key=None):
-    import subprocess
-    import time
-    import os
-    
-    # Set root password
-    subprocess.run(["bash", "-c", f"echo 'root:{ssh_password}' | chpasswd"], check=True)
-    
-    # Start SSH service
-    subprocess.run(["service", "ssh", "start"], check=True)
-    
-    # Setup environment
-    os.environ['PS1'] = r'\[\e[1;32m\]modal:\[\e[1;34m\]\w\[\e[0m\]$ '
-    
-    # Set OpenAI API key if provided
-    if openai_api_key:
-        os.environ['OPENAI_API_KEY'] = openai_api_key
-        print(f"✅ Set OpenAI API key in container environment (length: {len(openai_api_key)})")
-    else:
-        print("⚠️ No OpenAI API key provided to container")
-    
-    # Clone repository if provided
-    if repo_url:
-        repo_name_from_url = repo_name or repo_url.split('/')[-1].replace('.git', '')
-        print(f"📥 Cloning repository: {repo_url}")
-        
-        try:
-            subprocess.run(["git", "clone", repo_url], check=True, cwd="/root")
-            print(f"✅ Repository cloned successfully: {repo_name_from_url}")
-            
-            # Change to repository directory
-            repo_dir = f"/root/{repo_name_from_url}"
-            if os.path.exists(repo_dir):
-                os.chdir(repo_dir)
-                print(f"📂 Changed to repository directory: {repo_dir}")
-            
-        except subprocess.CalledProcessError as e:
-            print(f"❌ Failed to clone repository: {e}")
-    
-    # Run setup commands if provided
-    if setup_commands:
-        print(f"⚙️ Running {len(setup_commands)} setup commands...")
-        
-        # First, let's check the current directory structure
-        print("🔍 Checking current directory structure before running setup commands...")
-        try:
-            result = subprocess.run("pwd && ls -la", shell=True, check=True, 
-                                  capture_output=True, text=True)
-            print(f"📂 Current directory: {result.stdout}")
-        except subprocess.CalledProcessError as e:
-            print(f"⚠️ Could not check directory structure: {e}")
-        
-        # Define a simple run_command function for SSH container
-        def run_command_with_llm_debug(cmd, show_output=True, retry_count=0, max_retries=3):
-            """Execute a command with LLM debugging enabled"""
-            print(f"🔧 Executing: {cmd}")
-            try:
-                # Handle special case for source command which doesn't work with subprocess.run
-                if cmd.strip().startswith("source ") or " source " in cmd:
-                    print("⚠️ Detected 'source' command which doesn't work with subprocess.run")
-                    print("🔄 Converting to bash -c with dot (.) instead of source")
-                    # Replace source with . (dot) which is the same as source but works in sh
-                    modified_cmd = cmd.replace("source ", ". ")
-                    # Wrap in bash -c to ensure it runs in bash
-                    bash_cmd = f"bash -c '{modified_cmd}'"
-                    print(f"🔄 Modified command: {bash_cmd}")
-                    result = subprocess.run(bash_cmd, shell=True, check=True, 
-                                          capture_output=True, text=True)
-                else:
-                    result = subprocess.run(cmd, shell=True, check=True, 
-                                          capture_output=True, text=True)
-                
-                if result.stdout and show_output:
-                    print(f"✅ Output: {result.stdout}")
-                return True, result.stdout, ""
-            except subprocess.CalledProcessError as e:
-                error_output = e.stderr if e.stderr else str(e)
-                print(f"❌ Command failed: {e}")
-                print(f"❌ Error: {error_output}")
-                
-                # Call OpenAI for debugging
-                print("🔍 Attempting to debug the failed command with OpenAI...")
-                try:
-                    # Get the current directory for context
-                    current_dir = os.getcwd()
-                    
-                    # Call OpenAI for debugging
-                    print(f"🔍 DEBUG: About to call call_openai_for_debug...")
-                    print(f"🔍 DEBUG: Command: {cmd}")
-                    print(f"🔍 DEBUG: Error output length: {len(error_output)}")
-                    print(f"🔍 DEBUG: Current directory: {current_dir}")
-                    
-                    # Get the API key from environment or use the one that was fetched earlier
-                    api_key = os.environ.get("OPENAI_API_KEY")
-                    fix_command = call_openai_for_debug(cmd, error_output, api_key=api_key, current_dir=current_dir)
-                    
-                    print(f"🔍 DEBUG: call_openai_for_debug returned: {fix_command}")
-                    
-                    if fix_command:
-                        print(f"🔧 OpenAI suggested fix command: {fix_command}")
-                        
-                        # Run the fix command
-                        print(f"🔄 Running suggested fix command: {fix_command}")
-                        try:
-                            fix_result = subprocess.run(fix_command, shell=True, check=True, 
-                                                      capture_output=True, text=True)
-                            if fix_result.stdout:
-                                print(f"✅ Fix command output: {fix_result.stdout}")
-                            
-                            # Retry the original command
-                            print(f"🔄 Retrying original command: {cmd}")
-                            return run_command_with_llm_debug(cmd, show_output, retry_count + 1, max_retries)
-                        except subprocess.CalledProcessError as fix_e:
-                            print(f"❌ Fix command also failed: {fix_e}")
-                            return False, "", error_output
-                    else:
-                        print("❌ No fix suggested by OpenAI")
-                        return False, "", error_output
-                        
-                except Exception as debug_e:
-                    print(f"❌ LLM debugging failed: {debug_e}")
-                    return False, "", error_output
-        
-        for i, cmd in enumerate(setup_commands, 1):
-            print(f"📋 Executing command {i}/{len(setup_commands)}: {cmd}")
-            
-            # Check if this is a cd command and if the directory exists
-            if cmd.strip().startswith("cd "):
-                cd_parts = cmd.split(None, 1)
-                if len(cd_parts) >= 2:
-                    target_dir = cd_parts[1].strip('"\'')
-                    print(f"🔍 Checking if directory exists: {target_dir}")
-                    try:
-                        check_result = subprocess.run(f"test -d '{target_dir}'", shell=True, 
-                                                   capture_output=True, text=True)
-                        if check_result.returncode != 0:
-                            print(f"⚠️ Directory does not exist: {target_dir}")
-                            print(f"🔍 Current directory contents:")
-                            subprocess.run("pwd && ls -la", shell=True, check=False)
-                            
-                            # Try to find similar directories
-                            print(f"🔍 Looking for similar directories...")
-                            subprocess.run("find . -type d -name '*llama*' -o -name '*nano*' 2>/dev/null | head -10", shell=True, check=False)
-                    except Exception as e:
-                        print(f"⚠️ Could not check directory: {e}")
-            
-            success, stdout, stderr = run_command_with_llm_debug(cmd, show_output=True)
-            if not success:
-                print(f"⚠️ Command {i} failed, but continuing with remaining commands...")
-                
-                # If this was a cd command that failed, try to understand the directory structure
-                if cmd.strip().startswith("cd ") and "No such file or directory" in stderr:
-                    print(f"🔍 Analyzing directory structure after failed cd command...")
-                    subprocess.run("pwd && ls -la && echo '--- Parent directory ---' && ls -la ..", shell=True, check=False)
-    
-    # Get container info
-    print("🔍 Container started successfully!")
-    print(f"🆔 Container ID: {os.environ.get('MODAL_TASK_ID', 'unknown')}")
-    
-    # Keep the container running
-    while True:
-        time.sleep(30)
-        # Check if SSH service is still running
-        try:
-            subprocess.run(["service", "ssh", "status"], check=True, 
-                         capture_output=True)
-        except subprocess.CalledProcessError:
-            print("⚠️ SSH service stopped, restarting...")
-            subprocess.run(["service", "ssh", "start"], check=True)
 
 # Now modify the create_modal_ssh_container function to use the standalone ssh_container_function
 def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_commands=None, 
@@ -1400,7 +1188,7 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
             .apt_install(
                 "openssh-server", "sudo", "curl", "wget", "vim", "htop", "git", 
                 "python3", "python3-pip", "build-essential", "tmux", "screen", "nano",
-                "gpg", "ca-certificates", "software-properties-common"
+                "gpg", "ca-certificates", "software-properties-common", "nvtop"
             )
             .pip_install("uv", "modal", "requests", "openai")  # Fast Python package installer and Modal
             .run_commands(
@@ -1409,6 +1197,9 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
                 "mkdir -p /root/.ssh",
                 "chmod 700 /root/.ssh",
                 
+                # Generate SSH host keys
+                "ssh-keygen -A",
+                
                 # Configure SSH server
                 "sed -i 's/#PermitRootLogin prohibit-password/PermitRootLogin yes/' /etc/ssh/sshd_config",
                 "sed -i 's/#PasswordAuthentication yes/PasswordAuthentication yes/' /etc/ssh/sshd_config",
@@ -1418,8 +1209,9 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
                 "echo 'ClientAliveInterval 60' >> /etc/ssh/sshd_config",
                 "echo 'ClientAliveCountMax 3' >> /etc/ssh/sshd_config",
                 
-                # Generate SSH host keys
-                "ssh-keygen -A",
+                # Set up CUDA environment
+                "echo 'export PATH=/usr/local/cuda/bin:$PATH' >> /root/.bashrc",
+                "echo 'export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH' >> /root/.bashrc",
                 
                 # Set up a nice bash prompt
                 "echo 'export PS1=\"\\[\\e[1;32m\\]modal:\\[\\e[1;34m\\]\\w\\[\\e[0m\\]$ \"' >> /root/.bashrc",
@@ -1435,7 +1227,7 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
     if volume:
         volumes_config[volume_mount_path] = volume
         
-    # Define the SSH container function
+    # Define the SSH container function - simplified like the example
     @app.function(
         image=ssh_image,
         timeout=timeout_minutes * 60,  # Convert to seconds
@@ -1445,7 +1237,7 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
         serialized=True,
         volumes=volumes_config if volumes_config else None,
     )
-    def ssh_container_function(ssh_password=None, repo_url=None, repo_name=None, setup_commands=None, openai_api_key=None):
+    def start_ssh():
         """Start SSH container with password authentication and optional setup."""
         import subprocess
         import time
@@ -1455,14 +1247,25 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
         subprocess.run(["bash", "-c", f"echo 'root:{ssh_password}' | chpasswd"], check=True)
         
         # Set OpenAI API key if provided
+        openai_api_key = os.environ.get("OPENAI_API_KEY")
         if openai_api_key:
             os.environ['OPENAI_API_KEY'] = openai_api_key
             print(f"✅ Set OpenAI API key in container environment (length: {len(openai_api_key)})")
         else:
             print("⚠️ No OpenAI API key provided to container")
         
-        # Start SSH service
-        subprocess.run(["service", "ssh", "start"], check=True)
+        # Start SSH service using Popen (non-blocking) like in the example
+        subprocess.Popen(["/usr/sbin/sshd", "-D"])
+        time.sleep(2)  # Give SSH time to start
+        
+        # Test CUDA setup
+        try:
+            print("🔧 Testing CUDA setup...")
+            subprocess.run(["nvidia-smi"], check=True)
+            subprocess.run(["nvcc", "--version"], check=True)
+            print("✅ CUDA setup verified")
+        except subprocess.CalledProcessError as e:
+            print(f"⚠️ CUDA test failed: {e}")
         
         # Clone repository if provided
         if repo_url:
@@ -1590,43 +1393,28 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
                         print(f"🔍 Analyzing directory structure after failed cd command...")
                         subprocess.run("pwd && ls -la && echo '--- Parent directory ---' && ls -la ..", shell=True, check=False)
         
-        # Create SSH tunnel
-        with modal.forward(22, unencrypted=True) as tunnel:
-            host, port = tunnel.tcp_socket
+        # Forward SSH port and keep container alive - exactly like the example
+        with modal.forward(port=22, unencrypted=True) as tunnel:
+            hostname, port = tunnel.tcp_socket
             
             print("\n" + "=" * 80)
             print("🎉 SSH CONTAINER IS READY!")
             print("=" * 80)
-            print(f"🌐 SSH Host: {host}")
-            print(f"🔌 SSH Port: {port}")
-            print(f"👤 Username: root")
-            print(f"🔐 Password: {ssh_password}")
-            print()
-            print("🔗 CONNECT USING THIS COMMAND:")
-            print(f"ssh -p {port} root@{host}")
+            print(f"SSH: ssh -p {port} root@{hostname}")
+            print(f"Password: {ssh_password}")
             print("=" * 80)
             
-            # Keep the container running
+            # Keep alive - simplified like the example
             while True:
-                time.sleep(30)
-                # Check if SSH service is still running
-                try:
-                    subprocess.run(["service", "ssh", "status"], check=True, 
-                                 capture_output=True)
-                except subprocess.CalledProcessError:
-                    print("⚠️ SSH service stopped, restarting...")
-                    subprocess.run(["service", "ssh", "start"], check=True)
+                time.sleep(60)
     
     # Run the container
     try:
         print("⏳ Starting container... This may take 1-2 minutes...")
         
-        # Start the container in a new thread to avoid blocking
-        with modal.enable_output():
-            with app.run():
-                # Get the API key from environment
-                api_key = os.environ.get("OPENAI_API_KEY")
-                ssh_container_function.remote(ssh_password, repo_url, repo_name, setup_commands, api_key)
+        # Start the container - simplified like the example
+        with app.run():
+            start_ssh.remote()
                 
         # Clean up Modal token after container is successfully created
         cleanup_modal_token()

package/test_modalSandboxScript.py

@@ -968,218 +968,6 @@ def generate_random_password(length=16):
     password = ''.join(secrets.choice(alphabet) for i in range(length))
     return password
 
-# First, add the standalone ssh_container function at the module level, before the create_modal_ssh_container function
-
-# Define a module-level ssh container function
-ssh_app = modal.App("ssh-container-app")
-
-@ssh_app.function(
-    image=modal.Image.debian_slim()
-        .apt_install(
-            "openssh-server", "sudo", "curl", "wget", "vim", "htop", "git", 
-            "python3", "python3-pip", "build-essential", "tmux", "screen", "nano",
-            "gpg", "ca-certificates", "software-properties-common"
-        )
-        .pip_install("uv", "modal", "requests", "openai")  # Fast Python package installer and Modal
-        .run_commands(
-            # Create SSH directory
-            "mkdir -p /var/run/sshd",
-            "mkdir -p /root/.ssh",
-            "chmod 700 /root/.ssh",
-            
-            # Configure SSH server
-            "sed -i 's/#PermitRootLogin prohibit-password/PermitRootLogin yes/' /etc/ssh/sshd_config",
-            "sed -i 's/#PasswordAuthentication yes/PasswordAuthentication yes/' /etc/ssh/sshd_config",
-            "sed -i 's/#PubkeyAuthentication yes/PubkeyAuthentication yes/' /etc/ssh/sshd_config",
-            
-            # SSH keep-alive settings
-            "echo 'ClientAliveInterval 60' >> /etc/ssh/sshd_config",
-            "echo 'ClientAliveCountMax 3' >> /etc/ssh/sshd_config",
-            
-            # Generate SSH host keys
-            "ssh-keygen -A",
-            
-            # Install Modal CLI
-            "pip install modal",
-            
-            # Set up a nice bash prompt
-            "echo 'export PS1=\"\\[\\e[1;32m\\]modal:\\[\\e[1;34m\\]\\w\\[\\e[0m\\]$ \"' >> /root/.bashrc",
-        ),
-    timeout=3600,  # Default 1 hour timeout
-    gpu="a10g",    # Default GPU - this will be overridden when called
-    cpu=2,
-    memory=8192,
-    serialized=True,
-)
-def ssh_container_function(ssh_password, repo_url=None, repo_name=None, setup_commands=None, openai_api_key=None):
-    import subprocess
-    import time
-    import os
-    
-    # Set root password
-    subprocess.run(["bash", "-c", f"echo 'root:{ssh_password}' | chpasswd"], check=True)
-    
-    # Start SSH service
-    subprocess.run(["service", "ssh", "start"], check=True)
-    
-    # Setup environment
-    os.environ['PS1'] = r'\[\e[1;32m\]modal:\[\e[1;34m\]\w\[\e[0m\]$ '
-    
-    # Set OpenAI API key if provided
-    if openai_api_key:
-        os.environ['OPENAI_API_KEY'] = openai_api_key
-        print(f"✅ Set OpenAI API key in container environment (length: {len(openai_api_key)})")
-    else:
-        print("⚠️ No OpenAI API key provided to container")
-    
-    # Clone repository if provided
-    if repo_url:
-        repo_name_from_url = repo_name or repo_url.split('/')[-1].replace('.git', '')
-        print(f"📥 Cloning repository: {repo_url}")
-        
-        try:
-            subprocess.run(["git", "clone", repo_url], check=True, cwd="/root")
-            print(f"✅ Repository cloned successfully: {repo_name_from_url}")
-            
-            # Change to repository directory
-            repo_dir = f"/root/{repo_name_from_url}"
-            if os.path.exists(repo_dir):
-                os.chdir(repo_dir)
-                print(f"📂 Changed to repository directory: {repo_dir}")
-            
-        except subprocess.CalledProcessError as e:
-            print(f"❌ Failed to clone repository: {e}")
-    
-    # Run setup commands if provided
-    if setup_commands:
-        print(f"⚙️ Running {len(setup_commands)} setup commands...")
-        
-        # First, let's check the current directory structure
-        print("🔍 Checking current directory structure before running setup commands...")
-        try:
-            result = subprocess.run("pwd && ls -la", shell=True, check=True, 
-                                  capture_output=True, text=True)
-            print(f"📂 Current directory: {result.stdout}")
-        except subprocess.CalledProcessError as e:
-            print(f"⚠️ Could not check directory structure: {e}")
-        
-        # Define a simple run_command function for SSH container
-        def run_command_with_llm_debug(cmd, show_output=True, retry_count=0, max_retries=3):
-            """Execute a command with LLM debugging enabled"""
-            print(f"🔧 Executing: {cmd}")
-            try:
-                # Handle special case for source command which doesn't work with subprocess.run
-                if cmd.strip().startswith("source ") or " source " in cmd:
-                    print("⚠️ Detected 'source' command which doesn't work with subprocess.run")
-                    print("🔄 Converting to bash -c with dot (.) instead of source")
-                    # Replace source with . (dot) which is the same as source but works in sh
-                    modified_cmd = cmd.replace("source ", ". ")
-                    # Wrap in bash -c to ensure it runs in bash
-                    bash_cmd = f"bash -c '{modified_cmd}'"
-                    print(f"🔄 Modified command: {bash_cmd}")
-                    result = subprocess.run(bash_cmd, shell=True, check=True, 
-                                          capture_output=True, text=True)
-                else:
-                    result = subprocess.run(cmd, shell=True, check=True, 
-                                          capture_output=True, text=True)
-                
-                if result.stdout and show_output:
-                    print(f"✅ Output: {result.stdout}")
-                return True, result.stdout, ""
-            except subprocess.CalledProcessError as e:
-                error_output = e.stderr if e.stderr else str(e)
-                print(f"❌ Command failed: {e}")
-                print(f"❌ Error: {error_output}")
-                
-                # Call OpenAI for debugging
-                print("🔍 Attempting to debug the failed command with OpenAI...")
-                try:
-                    # Get the current directory for context
-                    current_dir = os.getcwd()
-                    
-                    # Call OpenAI for debugging
-                    print(f"🔍 DEBUG: About to call call_openai_for_debug...")
-                    print(f"🔍 DEBUG: Command: {cmd}")
-                    print(f"🔍 DEBUG: Error output length: {len(error_output)}")
-                    print(f"🔍 DEBUG: Current directory: {current_dir}")
-                    
-                    # Get the API key from environment or use the one that was fetched earlier
-                    api_key = os.environ.get("OPENAI_API_KEY")
-                    fix_command = call_openai_for_debug(cmd, error_output, api_key=api_key, current_dir=current_dir)
-                    
-                    print(f"🔍 DEBUG: call_openai_for_debug returned: {fix_command}")
-                    
-                    if fix_command:
-                        print(f"🔧 OpenAI suggested fix command: {fix_command}")
-                        
-                        # Run the fix command
-                        print(f"🔄 Running suggested fix command: {fix_command}")
-                        try:
-                            fix_result = subprocess.run(fix_command, shell=True, check=True, 
-                                                      capture_output=True, text=True)
-                            if fix_result.stdout:
-                                print(f"✅ Fix command output: {fix_result.stdout}")
-                            
-                            # Retry the original command
-                            print(f"🔄 Retrying original command: {cmd}")
-                            return run_command_with_llm_debug(cmd, show_output, retry_count + 1, max_retries)
-                        except subprocess.CalledProcessError as fix_e:
-                            print(f"❌ Fix command also failed: {fix_e}")
-                            return False, "", error_output
-                    else:
-                        print("❌ No fix suggested by OpenAI")
-                        return False, "", error_output
-                        
-                except Exception as debug_e:
-                    print(f"❌ LLM debugging failed: {debug_e}")
-                    return False, "", error_output
-        
-        for i, cmd in enumerate(setup_commands, 1):
-            print(f"📋 Executing command {i}/{len(setup_commands)}: {cmd}")
-            
-            # Check if this is a cd command and if the directory exists
-            if cmd.strip().startswith("cd "):
-                cd_parts = cmd.split(None, 1)
-                if len(cd_parts) >= 2:
-                    target_dir = cd_parts[1].strip('"\'')
-                    print(f"🔍 Checking if directory exists: {target_dir}")
-                    try:
-                        check_result = subprocess.run(f"test -d '{target_dir}'", shell=True, 
-                                                   capture_output=True, text=True)
-                        if check_result.returncode != 0:
-                            print(f"⚠️ Directory does not exist: {target_dir}")
-                            print(f"🔍 Current directory contents:")
-                            subprocess.run("pwd && ls -la", shell=True, check=False)
-                            
-                            # Try to find similar directories
-                            print(f"🔍 Looking for similar directories...")
-                            subprocess.run("find . -type d -name '*llama*' -o -name '*nano*' 2>/dev/null | head -10", shell=True, check=False)
-                    except Exception as e:
-                        print(f"⚠️ Could not check directory: {e}")
-            
-            success, stdout, stderr = run_command_with_llm_debug(cmd, show_output=True)
-            if not success:
-                print(f"⚠️ Command {i} failed, but continuing with remaining commands...")
-                
-                # If this was a cd command that failed, try to understand the directory structure
-                if cmd.strip().startswith("cd ") and "No such file or directory" in stderr:
-                    print(f"🔍 Analyzing directory structure after failed cd command...")
-                    subprocess.run("pwd && ls -la && echo '--- Parent directory ---' && ls -la ..", shell=True, check=False)
-    
-    # Get container info
-    print("🔍 Container started successfully!")
-    print(f"🆔 Container ID: {os.environ.get('MODAL_TASK_ID', 'unknown')}")
-    
-    # Keep the container running
-    while True:
-        time.sleep(30)
-        # Check if SSH service is still running
-        try:
-            subprocess.run(["service", "ssh", "status"], check=True, 
-                         capture_output=True)
-        except subprocess.CalledProcessError:
-            print("⚠️ SSH service stopped, restarting...")
-            subprocess.run(["service", "ssh", "start"], check=True)
 
 # Now modify the create_modal_ssh_container function to use the standalone ssh_container_function
 def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_commands=None, 
@@ -1400,7 +1188,7 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
             .apt_install(
                 "openssh-server", "sudo", "curl", "wget", "vim", "htop", "git", 
                 "python3", "python3-pip", "build-essential", "tmux", "screen", "nano",
-                "gpg", "ca-certificates", "software-properties-common"
+                "gpg", "ca-certificates", "software-properties-common", "nvtop"
             )
             .pip_install("uv", "modal", "requests", "openai")  # Fast Python package installer and Modal
             .run_commands(
@@ -1409,6 +1197,9 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
                 "mkdir -p /root/.ssh",
                 "chmod 700 /root/.ssh",
                 
+                # Generate SSH host keys
+                "ssh-keygen -A",
+                
                 # Configure SSH server
                 "sed -i 's/#PermitRootLogin prohibit-password/PermitRootLogin yes/' /etc/ssh/sshd_config",
                 "sed -i 's/#PasswordAuthentication yes/PasswordAuthentication yes/' /etc/ssh/sshd_config",
@@ -1418,8 +1209,9 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
                 "echo 'ClientAliveInterval 60' >> /etc/ssh/sshd_config",
                 "echo 'ClientAliveCountMax 3' >> /etc/ssh/sshd_config",
                 
-                # Generate SSH host keys
-                "ssh-keygen -A",
+                # Set up CUDA environment
+                "echo 'export PATH=/usr/local/cuda/bin:$PATH' >> /root/.bashrc",
+                "echo 'export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH' >> /root/.bashrc",
                 
                 # Set up a nice bash prompt
                 "echo 'export PS1=\"\\[\\e[1;32m\\]modal:\\[\\e[1;34m\\]\\w\\[\\e[0m\\]$ \"' >> /root/.bashrc",
@@ -1435,7 +1227,7 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
     if volume:
         volumes_config[volume_mount_path] = volume
         
-    # Define the SSH container function
+    # Define the SSH container function - simplified like the example
     @app.function(
         image=ssh_image,
         timeout=timeout_minutes * 60,  # Convert to seconds
@@ -1445,7 +1237,7 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
         serialized=True,
         volumes=volumes_config if volumes_config else None,
     )
-    def ssh_container_function(ssh_password=None, repo_url=None, repo_name=None, setup_commands=None, openai_api_key=None):
+    def start_ssh():
         """Start SSH container with password authentication and optional setup."""
         import subprocess
         import time
@@ -1455,14 +1247,25 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
         subprocess.run(["bash", "-c", f"echo 'root:{ssh_password}' | chpasswd"], check=True)
         
         # Set OpenAI API key if provided
+        openai_api_key = os.environ.get("OPENAI_API_KEY")
         if openai_api_key:
             os.environ['OPENAI_API_KEY'] = openai_api_key
             print(f"✅ Set OpenAI API key in container environment (length: {len(openai_api_key)})")
         else:
             print("⚠️ No OpenAI API key provided to container")
         
-        # Start SSH service
-        subprocess.run(["service", "ssh", "start"], check=True)
+        # Start SSH service using Popen (non-blocking) like in the example
+        subprocess.Popen(["/usr/sbin/sshd", "-D"])
+        time.sleep(2)  # Give SSH time to start
+        
+        # Test CUDA setup
+        try:
+            print("🔧 Testing CUDA setup...")
+            subprocess.run(["nvidia-smi"], check=True)
+            subprocess.run(["nvcc", "--version"], check=True)
+            print("✅ CUDA setup verified")
+        except subprocess.CalledProcessError as e:
+            print(f"⚠️ CUDA test failed: {e}")
         
         # Clone repository if provided
         if repo_url:
@@ -1590,43 +1393,28 @@ def create_modal_ssh_container(gpu_type, repo_url=None, repo_name=None, setup_co
                         print(f"🔍 Analyzing directory structure after failed cd command...")
                         subprocess.run("pwd && ls -la && echo '--- Parent directory ---' && ls -la ..", shell=True, check=False)
         
-        # Create SSH tunnel
-        with modal.forward(22, unencrypted=True) as tunnel:
-            host, port = tunnel.tcp_socket
+        # Forward SSH port and keep container alive - exactly like the example
+        with modal.forward(port=22, unencrypted=True) as tunnel:
+            hostname, port = tunnel.tcp_socket
             
             print("\n" + "=" * 80)
             print("🎉 SSH CONTAINER IS READY!")
             print("=" * 80)
-            print(f"🌐 SSH Host: {host}")
-            print(f"🔌 SSH Port: {port}")
-            print(f"👤 Username: root")
-            print(f"🔐 Password: {ssh_password}")
-            print()
-            print("🔗 CONNECT USING THIS COMMAND:")
-            print(f"ssh -p {port} root@{host}")
+            print(f"SSH: ssh -p {port} root@{hostname}")
+            print(f"Password: {ssh_password}")
             print("=" * 80)
             
-            # Keep the container running
+            # Keep alive - simplified like the example
             while True:
-                time.sleep(30)
-                # Check if SSH service is still running
-                try:
-                    subprocess.run(["service", "ssh", "status"], check=True, 
-                                 capture_output=True)
-                except subprocess.CalledProcessError:
-                    print("⚠️ SSH service stopped, restarting...")
-                    subprocess.run(["service", "ssh", "start"], check=True)
+                time.sleep(60)
     
     # Run the container
     try:
         print("⏳ Starting container... This may take 1-2 minutes...")
         
-        # Start the container in a new thread to avoid blocking
-        with modal.enable_output():
-            with app.run():
-                # Get the API key from environment
-                api_key = os.environ.get("OPENAI_API_KEY")
-                ssh_container_function.remote(ssh_password, repo_url, repo_name, setup_commands, api_key)
+        # Start the container - simplified like the example
+        with app.run():
+            start_ssh.remote()
                 
         # Clean up Modal token after container is successfully created
         cleanup_modal_token()