BeagleBone Black(rev.C) Debian下Python程序运行数小时后停止求助

Hey there, sorry to hear your data logging program is cutting out after 9 hours of runtime—let’s walk through practical troubleshooting steps and optimization ideas to get this sorted out:

The first thing to do is figure out why it’s stopping—without logs, you’re guessing. Wrap your core logic in try/except blocks and write detailed logs to a file, including stack traces for crashes.

Example code snippet:

import logging
from datetime import datetime

# Configure logging to write to a file with timestamps and error details
logging.basicConfig(
    filename='logger_errors.log',
    level=logging.DEBUG,
    format='%(asctime)s | %(levelname)s | %(message)s'
)

try:
    for task_num in range(1, 10):  # Your 9-hour task loop
        logging.info(f"Starting task {task_num}/9")
        # --- Your data recording logic here ---
        logging.info(f"Completed task {task_num}/9 successfully")
        # Add your 1-hour delay logic here (e.g., time.sleep(3600))
except Exception as e:
    # Capture full error stack trace for debugging
    logging.exception(f"Fatal error during task {task_num}: {str(e)}")

After the program stops, check logger_errors.log—the last entries will tell you if it crashed on a specific task, hit an I/O error, or ran into a hardware issue.

BeagleBone Black has limited RAM (~512MB), and Python’s garbage collector doesn’t always catch everything. If your program accumulates data in global variables, leaves file handles open, or creates unclosed objects, it might get killed by the system’s OOM (Out-of-Memory) killer.

• Monitor memory in-code: Use the tracemalloc module to track memory usage over time:

  import tracemalloc
  tracemalloc.start()
  
  # Run your task loop here...
  
  # Take a memory snapshot after a few tasks
  snapshot = tracemalloc.take_snapshot()
  top_memory_hogs = snapshot.statistics('lineno')
  logging.info("\nTop 10 memory-consuming lines:")
  for stat in top_memory_hogs[:10]:
      logging.info(stat)
  

• System-level check: Run htop in a separate PuTTY session while your program runs. Watch the Python process’s RES (resident memory) column—if it keeps climbing steadily, you’ve got a leak to fix (e.g., use with open(...) as f: to auto-close files, clear unused lists/dictionaries after each task).

The Linux kernel might terminate your program if it’s using too many resources. Check system logs for clues:

• Run dmesg | grep -i oom to see if the OOM killer targeted your Python process.
• Check system logs with cat /var/log/syslog | grep -i python for any kernel messages about stopping your program.

Double-check your task loop—if you hardcoded range(9) for 9 tasks, the program will stop intentionally after 9 runs! If you want it to run indefinitely, switch to a while True: loop with a check to stop on demand (e.g., a flag file or keyboard interrupt).

If your program interacts with sensors/peripherals (e.g., ADC, I2C devices), flaky hardware connections or unhandled timeouts could cause it to hang:

• Add timeout parameters to hardware read/write calls (e.g., serial.Serial(..., timeout=5) for UART).
• Use signal to implement a watchdog timer that restarts a stuck task:

  import signal
  
  def timeout_handler(signum, frame):
      raise TimeoutError("Hardware read timed out")
  
  # Set a 30-second timeout for hardware operations
  signal.signal(signal.SIGALRM, timeout_handler)
  signal.alarm(30)
  
  # Your hardware read logic here
  signal.alarm(0)  # Disable timeout after successful read
  

If you’re running the program directly in PuTTY, a lost SSH connection could terminate it (even if the BBB is powered). Use systemd to run it as a persistent service that auto-restarts on crashes:

1. Create a service file at /etc/systemd/system/data-logger.service:

   [Unit]
   Description=BeagleBone Data Logger
   After=multi-user.target
   
   [Service]
   Type=simple
   User=debian
   ExecStart=/usr/bin/python3 /home/debian/your_logger_program.py
   Restart=always
   RestartSec=10
   StandardOutput=journal+console
   StandardError=journal+console
   
   [Install]
   WantedBy=multi-user.target
   

2. Enable and start the service:

   sudo systemctl daemon-reload
   sudo systemctl enable data-logger.service
   sudo systemctl start data-logger.service
   

3. Check service logs anytime with:

   journalctl -u data-logger.service -f
   

内容的提问来源于stack exchange，提问作者Nguyen Hiep