通过X-CTU控制脚本启停遇问题：adxl345test.py无法被终止求助

Let's break down why your current setup isn't stopping the data collection script, and walk through fixes to get it working reliably.

Key Issues in Your Current Code #

1. shell=True creates an extra process layer: When you use shell=True in subprocess.Popen, you're spawning a shell that runs your Python script—not the script directly. So the p variable refers to the shell, not the actual adxl345test.py process. Killing the shell leaves the script running as an orphan process.
2. Serial port reinitialized every loop: Opening a new serial connection on every iteration is inefficient and can cause connection glitches.
3. Uninitialized p variable: If you send "stop" before "run", your script will throw a NameError because p isn't defined.
4. No process state checks: You don't verify if the script is already running before starting it, or if it's still active before trying to stop it.

Revised Working Code #

Here's a fixed version of zb_control.py that addresses all these issues:

import serial
import time
import subprocess
import os
import signal

# Initialize serial connection once (outside loop to avoid repeated openings)
try:
    ser = serial.Serial('/dev/ttyUSB0', 9600, timeout=2)
    print("Serial connection established successfully")
except serial.SerialException as e:
    print(f"Failed to open serial port: {e}")
    exit(1)

# Track the data collection process (initialize to None)
data_process = None

while True:
    # Read and decode serial input (convert bytes to string)
    incoming_cmd = ser.readline().strip().decode('utf-8')
    if not incoming_cmd:
        continue  # Skip empty lines from serial
    
    if incoming_cmd == 'run':
        # Only start if no active process exists
        if data_process is None or data_process.poll() is not None:
            print("---------------")
            print("Collecting data")
            print("---------------")
            # Spawn the script directly (no shell=True) to target the actual process
            data_process = subprocess.Popen(
                ["python3", "/home/pi/adxl345test.py"],
                stdout=subprocess.PIPE,
                stderr=subprocess.PIPE,
                start_new_session=True  # Optional: for killing child processes too
            )
        else:
            print("Data collection is already running")
    
    elif incoming_cmd == 'stop':
        # Only stop if the process is active
        if data_process is not None and data_process.poll() is None:
            try:
                # First try graceful termination (SIGTERM)
                os.killpg(os.getpgid(data_process.pid), signal.SIGTERM)
                data_process.wait(timeout=2)  # Wait for script to exit cleanly
                print("-----------------------")
                print("Script stopped gracefully")
                print("-----------------------")
            except subprocess.TimeoutExpired:
                # Force kill if graceful stop fails (SIGKILL)
                os.killpg(os.getpgid(data_process.pid), signal.SIGKILL)
                print("-----------------------")
                print("Script force-killed")
                print("-----------------------")
            finally:
                data_process = None  # Reset process tracker
        else:
            print("No active data collection script to stop")
    
    # Small delay to reduce CPU usage
    time.sleep(0.1)

Critical Fixes Explained #

1. Avoid shell=True: By using ["python3", "/home/pi/adxl345test.py"], we directly spawn the Python process. Now data_process refers to the actual script, so killing it stops the data collection.
2. Process state tracking: We check data_process.poll() (returns None if the process is running) to prevent duplicate starts and invalid stop attempts.
3. Graceful + force termination: We first send SIGTERM to let the script clean up (e.g., close the CSV file properly), then use SIGKILL if it doesn't exit within 2 seconds.
4. Process group killing: The start_new_session=True flag creates a new process group, so os.killpg kills the script and any child processes it spawns (useful if your data collection script uses subprocesses).

Additional Verification Steps #

• After sending "run", check the process ID with ps aux | grep adxl345test.py to confirm it's running.
• After sending "stop", re-run the above command to ensure the PID is gone.
• If the CSV file still appears to grow immediately after stopping, wait a few seconds—this is often OS disk caching, not the script still running.

内容的提问来源于stack exchange，提问作者user75374