interval cron issue with microseconds

I’m trying to setup a simple job with an interval of 5 minutes:

scheduler.add_job(
    func=ping_db,
    id='ping_db',
    trigger="interval",
    minutes=5,
    misfire_grace_time=None,
    replace_existing=True
)

note: happens with trigger='cron' too which is setup at 1am to run a process.

Expected Behavior

Since the function just ping a db with a sql statement select1. Its execution is really fast. I’m expecting to get in my logs:

INFO 2019-12-24 10:29:01,743 {file name: base.py Line: 144 -- run_job()} Job "ping_db (trigger: interval[0:05:00], next run at: 2019-12-24 18:34:00 UTC)" executed successfully
DEBUG 2019-12-24 10:29:01,767 {file name: base.py Line: 1020 -- _process_jobs()} Next wakeup is due at 2019-12-24 18:34:00+00:00 (in 299.406608 seconds)

Current Behavior

What I’m getting is multiple next run times and the only difference is the microseconds:

DEBUG 2019-12-24 10:29:01,767 {file name: base.py Line: 1020 -- _process_jobs()} Next wakeup is due at 2019-12-24 18:34:00+00:00 (in 299.406608 seconds)
DEBUG 2019-12-24 10:29:01,846 {file name: base.py Line: 1020 -- _process_jobs()} Next wakeup is due at 2019-12-24 18:34:00+00:00 (in 299.400619 seconds)
DEBUG 2019-12-24 10:29:01,928 {file name: base.py Line: 1020 -- _process_jobs()} Next wakeup is due at 2019-12-24 18:34:00+00:00 (in 299.390624 seconds)
DEBUG 2019-12-24 10:29:02,017 {file name: base.py Line: 1020 -- _process_jobs()} Next wakeup is due at 2019-12-24 18:34:00+00:00 (in 299.389619 seconds)
DEBUG 2019-12-24 10:29:02,354 {file name: base.py Line: 1020 -- _process_jobs()} Next wakeup is due at 2019-12-24 18:34:00+00:00 (in 298.807571 seconds)
DEBUG 2019-12-24 10:29:02,434 {file name: base.py Line: 1020 -- _process_jobs()} Next wakeup is due at 2019-12-24 18:34:00+00:00 (in 298.801572 seconds)

Application in Flask

Context (Environment)

Python 3.7.0 Windows 10 Microsoft Windows Server 1607 Internet Information Services 10.0.14393.0

I’m using just 1 worker in IIS, and configuring the scheduler in the app with:

    # app scheduler config
    app.config['SCHEDULER_JOBSTORES'] = {
        'default': SQLAlchemyJobStore(engine=engine)
    }
    app.config['SCHEDULER_EXECUTORS'] = {
        'default': {'type': 'threadpool', 'max_workers': 20}, # ThreadPoolExecutor(20),
        # 'processpool': ProcessPoolExecutor(5)
    }
    app.config['SCHEDULER_JOB_DEFAULTS'] = {
        'misfire_grace_time': None,
        'coalesce': True,
        'max_instances': 1,
        'replace_existing': True,
    }

tried 'coalesce': True and 'coalesce': False.

[agronholm]

You can either follow the recommended procedure for starting the scheduler in a separate process, or wait for APScheduler v4.0 which supports job store sharing.

[virtualdvid]

Hi @Laurel-rao sorry for my late replay. I tried what you mentioned a while ago and the behavior was the same. The application is now on production and the issue is less. It happens sometimes randomly. Fortunately this doesn’t affect the overall of the app.

@agronholm Great news, looking forward to see what’s coming in APScheduler v4.0!