Merge branch 'main' into eval_api_final

llama_stack/distribution/build.py

@@ -6,7 +6,6 @@
 
 import importlib.resources
 import logging
-import sys
 from pathlib import Path
 from typing import Dict, List
 
@@ -15,7 +14,7 @@ from termcolor import cprint
 
 from llama_stack.distribution.datatypes import BuildConfig, Provider
 from llama_stack.distribution.distribution import get_provider_registry
-from llama_stack.distribution.utils.exec import run_command, run_with_pty
+from llama_stack.distribution.utils.exec import run_command
 from llama_stack.distribution.utils.image_types import LlamaStackImageType
 from llama_stack.providers.datatypes import Api
 
@@ -123,11 +122,7 @@ def build_image(
     if special_deps:
         args.append("#".join(special_deps))
 
-    is_terminal = sys.stdin.isatty()
-    if is_terminal:
-        return_code = run_with_pty(args)
-    else:
-        return_code = run_command(args)
+    return_code = run_command(args)
 
     if return_code != 0:
         log.error(

llama_stack/distribution/build_container.sh

@@ -43,7 +43,7 @@ RED='\033[0;31m'
 NC='\033[0m' # No Color
 
 CONTAINER_BINARY=${CONTAINER_BINARY:-docker}
-CONTAINER_OPTS=${CONTAINER_OPTS:-}
+CONTAINER_OPTS=${CONTAINER_OPTS:---progress=plain}
 
 TEMP_DIR=$(mktemp -d)
 
@@ -253,8 +253,7 @@ $CONTAINER_BINARY build \
   "${CLI_ARGS[@]}" \
   -t "$image_tag" \
   -f "$TEMP_DIR/Containerfile" \
-  "." \
-  --progress=plain
+  "."
 
 # clean up tmp/configs
 set +x

llama_stack/distribution/providers.py

@@ -8,10 +8,13 @@
 from pydantic import BaseModel
 
 from llama_stack.apis.providers import ListProvidersResponse, ProviderInfo, Providers
+from llama_stack.log import get_logger
 
 from .datatypes import StackRunConfig
 from .stack import redact_sensitive_fields
 
+logger = get_logger(name=__name__, category="core")
+
 
 class ProviderImplConfig(BaseModel):
     run_config: StackRunConfig
@@ -31,6 +34,10 @@ class ProviderImpl(Providers):
     async def initialize(self) -> None:
         pass
 
+    async def shutdown(self) -> None:
+        logger.debug("ProviderImpl.shutdown")
+        pass
+
     async def list_providers(self) -> ListProvidersResponse:
         run_config = self.config.run_config
         safe_config = StackRunConfig(**redact_sensitive_fields(run_config.model_dump()))

llama_stack/distribution/utils/exec.py

@@ -4,13 +4,10 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-import errno
 import logging
 import os
-import select
 import signal
 import subprocess
-import sys
 
 from termcolor import cprint
 
@@ -88,13 +85,6 @@ def formulate_run_args(image_type, image_name, config, template_name) -> list:
     return run_args
 
 
-def run_with_pty(command):
-    if sys.platform.startswith("win"):
-        return _run_with_pty_win(command)
-    else:
-        return _run_with_pty_unix(command)
-
-
 def in_notebook():
     try:
         from IPython import get_ipython
@@ -108,19 +98,19 @@ def in_notebook():
     return True
 
 
-# run a command in a pseudo-terminal, with interrupt handling,
-# useful when you want to run interactive things
-def _run_with_pty_unix(command):
-    import pty
-    import termios
+def run_command(command: list[str]) -> int:
+    """
+    Run a command with interrupt handling and output capture.
+    Uses subprocess.run with direct stream piping for better performance.
 
-    master, slave = pty.openpty()
+    Args:
+        command (list): The command to run.
 
-    old_settings = termios.tcgetattr(sys.stdin)
+    Returns:
+        int: The return code of the command.
+    """
     original_sigint = signal.getsignal(signal.SIGINT)
-
     ctrl_c_pressed = False
-    process = None
 
     def sigint_handler(signum, frame):
         nonlocal ctrl_c_pressed
@@ -131,106 +121,19 @@ def _run_with_pty_unix(command):
         # Set up the signal handler
         signal.signal(signal.SIGINT, sigint_handler)
 
-        new_settings = termios.tcgetattr(sys.stdin)
-        new_settings[3] = new_settings[3] & ~termios.ECHO  # Disable echo
-        new_settings[3] = new_settings[3] & ~termios.ICANON  # Disable canonical mode
-        termios.tcsetattr(sys.stdin, termios.TCSADRAIN, new_settings)
-
-        process = subprocess.Popen(
+        # Run the command with stdout/stderr piped directly to system streams
+        result = subprocess.run(
             command,
-            stdin=slave,
-            stdout=slave,
-            stderr=slave,
-            universal_newlines=True,
-            preexec_fn=os.setsid,
+            text=True,
+            check=False,
         )
-
-        # Close the slave file descriptor as it's now owned by the subprocess
-        os.close(slave)
-
-        def handle_io():
-            while not ctrl_c_pressed:
-                try:
-                    rlist, _, _ = select.select([sys.stdin, master], [], [], 0.1)
-
-                    if sys.stdin in rlist:
-                        data = os.read(sys.stdin.fileno(), 1024)
-                        if not data:
-                            break
-                        os.write(master, data)
-
-                    if master in rlist:
-                        data = os.read(master, 1024)
-                        if not data:
-                            break
-                        sys.stdout.buffer.write(data)
-                        sys.stdout.flush()
-
-                except KeyboardInterrupt:
-                    # This will be raised when Ctrl+C is pressed
-                    break
-
-                if process.poll() is not None:
-                    break
-
-        handle_io()
-    except (EOFError, KeyboardInterrupt):
-        pass
-    except OSError as e:
-        if e.errno != errno.EIO:
-            raise
-    finally:
-        # Clean up
-        termios.tcsetattr(sys.stdin, termios.TCSADRAIN, old_settings)
-        signal.signal(signal.SIGINT, original_sigint)
-
-        os.close(master)
-        if process and process.poll() is None:
-            process.terminate()
-            process.wait()
-
-    return process.returncode
-
-
-# run a command in a pseudo-terminal in windows, with interrupt handling,
-def _run_with_pty_win(command):
-    """
-    Runs a command with interactive support using subprocess directly.
-    """
-    try:
-        # For shell scripts on Windows, use appropriate shell
-        if isinstance(command, (list, tuple)):
-            if command[0].endswith(".sh"):
-                if os.path.exists("/usr/bin/bash"):  # WSL
-                    command = ["bash"] + command
-                else:
-                    # Use cmd.exe with bash while preserving all arguments
-                    command = ["cmd.exe", "/c", "bash"] + command
-
-        process = subprocess.Popen(
-            command,
-            shell=True,
-            universal_newlines=True,
-        )
-
-        process.wait()
-
+        return result.returncode
+    except subprocess.SubprocessError as e:
+        log.error(f"Subprocess error: {e}")
+        return 1
     except Exception as e:
-        print(f"Error: {str(e)}")
+        log.exception(f"Unexpected error: {e}")
         return 1
     finally:
-        if process and process.poll() is None:
-            process.terminate()
-            process.wait()
-    return process.returncode
-
-
-def run_command(command):
-    try:
-        result = subprocess.run(command, capture_output=True, text=True, check=True)
-        print("Script Output\n", result.stdout)
-        return result.returncode
-    except subprocess.CalledProcessError as e:
-        print("Error running script:", e)
-        print("Error output:", e.stderr)
-        return e.returncode
+        # Restore the original signal handler
+        signal.signal(signal.SIGINT, original_sigint)