The task will cache?

Expected Behavior

Changes as database data changes

Current Behavior

If only one thread is opened, the data will never change. When I open 20 threads, the data of the thread after the database data changes is updated.

Steps to Reproduce

1. I use the Flask-apscheduler.
2. Use SQLAlchemy to get database data.
3. Create a task that runs every second to get database data.
4. Chang the data in database.

Context (Environment)

Win10 1903 python3.7 apscheduler3.6.3

Detailed Description

task code

def check_node_health():
    for node in Node.query.all():
        if node.records:
            last_time = datetime.datetime.min
            print(node.records)
            for record in Record.query.filter_by(chip_id=node.chip_id).all():
                last_time = max(last_time, record.create_datetime)
            print(last_time)
            print((datetime.datetime.now() - last_time).seconds)

config code

SCHEDULER_API_ENABLED = True
JOBS = [
    {
        'id': 'node_health_checker',
        'func': check_node_health,
        'args': '',
        'trigger': 'interval',
        'seconds': 1
    }
]
SCHEDULER_EXECUTORS = {
    'default': ThreadPoolExecutor(20),
}

As the picture shows:

The last line of data is what I added at runtime Duplicate tasks run by the thread, showing updated data,but some are not. So I suspect that the thread executing the task will cache the data.

[agronholm]

Your problem does not seem to have anything to do with scheduling. APScheduler does not set up or operate the database connection you’re using, so why have you reported this as an APScheduler problem? For the record, I think your problem stems from transaction isolation. You might not be closing transactions properly.

[0x587]

Thanks for your guidance, I found the problem.Thanks!!!