Django 1.8.1连接MSSQL执行长脚本时服务器频繁重启求助

Hey Steven, I’ve run into this exact head-scratcher before—dev server restarting silently halfway through a batch operation is so frustrating. Let’s break down the most likely causes and fixes tailored to your setup:

1. Disable Auto-Reload (The Quick Test) #

Django’s runserver has an auto-reload feature that watches for file changes in your project directory to restart the server. If your script is generating temporary files, writing logs, or modifying any files within the monitored paths (even accidentally), it’ll trigger a restart mid-execution.

Fix:
Run the server with auto-reload disabled to rule this out:

python manage.py runserver --noreload

If the script completes without restarting, you’ll need to either move your temp/log files outside the project directory or configure Django to ignore those paths. For Django 1.8, you can tweak the WATCH_FILES setting to exclude specific directories, but keeping --noreload enabled during batch testing is the simplest workaround.

2. Optimize Your Insert Logic (Fix Memory/Performance Issues) #

逐条插入250 records might not sound like much, but if each iteration pulls data, creates model instances, and hits the database individually, it can spike memory usage or cause long-running connections that trigger timeouts. Older Django versions (like 1.8) and MSSQL drivers can be sensitive to this kind of repeated overhead.

Fix: Use Bulk Inserts
Replace your loop of Model.objects.create(**data) with bulk_create, which sends all inserts in a single database query—way faster and easier on memory:

# Collect all processed data first
records_to_insert = []
for item in external_data_source:
    # Process your data as you normally would
    record = YourModel(
        field1=item['field1'],
        field2=item['field2'],
        # ... map other fields
    )
    records_to_insert.append(record)

# Bulk insert all records at once
YourModel.objects.bulk_create(records_to_insert)

Note: bulk_create in Django 1.8 doesn’t trigger pre_save/post_save signals, so if your model relies on those, you’ll need to handle them manually. But for raw data insertion, this is a massive improvement.

3. Check for MSSQL Connection Timeouts #

Django 1.8.1 is pretty old, and if you’re using an outdated version of pyodbc or the MSSQL driver, long-running scripts might hit connection timeouts that cause the server to restart silently.

Fixes:

• Update your MSSQL driver (like pyodbc) to the latest version compatible with Python 2.7/3.4 (since Django 1.8 supports these versions).
• Add extended connection timeout settings to your DATABASES config in settings.py:

DATABASES = {
    'default': {
        'ENGINE': 'sql_server.pyodbc',
        'NAME': 'your_db_name',
        'USER': 'your_username',
        'PASSWORD': 'your_password',
        'HOST': 'your_server_address',
        'PORT': '',
        'OPTIONS': {
            'driver': 'SQL Server Native Client 11.0',
            'connect_timeout': 300,  # 5 minutes, adjust as needed
        },
    }
}

4. Test with a Production Server (Rule Out Dev Server Limitations) #

The Django dev server (runserver) is built for quick development tasks—it’s single-threaded and not designed to handle long-running requests. Sometimes it’ll crash or restart under load without logging errors.

Fix:
Test your script using a lightweight production server like Gunicorn:

# Install a compatible Gunicorn version
pip install gunicorn==19.9.0  # Works well with Django 1.8
# Run the server
gunicorn your_project.wsgi:application

If the script runs without issues here, the problem is definitely the dev server’s limitations, and you should stick to using a proper server for batch operations.

Hope one of these fixes gets your script running smoothly!

内容的提问来源于stack exchange，提问作者Steven Tran