Race condition when adding a job once scheduler is started

[agronholm]

Originally reported by: chemicalstorm (Bitbucket: chemicalstorm, GitHub: chemicalstorm)

---

Using the following code:

#!python
from apscheduler.scheduler import Scheduler
logging.basicConfig(level=logging.DEBUG)

def job_function():
    print "Hello World"
    time.sleep(10)

# Start the scheduler
sched = Scheduler()

sched.start()
sched.add_cron_job(job_function, **{'minute':'*/2'})

It seems we can fall in a race condition. Several launches of this code display different outputs (I just added a log trace in scheduler.py, function //_real_add_job//)

Functionnal:

guestdev@guestdev-laptop:~/Projet/bac_a_sable/APScheduler-2.0.1$ python test_aps.py 
INFO:apscheduler.threadpool:Started thread pool with 0 core threads and 20 maximum threads
INFO:apscheduler.scheduler:Scheduler started
DEBUG:apscheduler.scheduler:Looking for jobs to run
DEBUG:apscheduler.scheduler:No jobs; waiting until a job is added
DEBUG:apscheduler.scheduler:Adding real job with wakeup: True
INFO:apscheduler.scheduler:Added job "job_function (trigger: cron[minute='*/2'], next run at: 2011-06-20 13:32:00)" to job store "default"
DEBUG:apscheduler.scheduler:Looking for jobs to run
DEBUG:apscheduler.scheduler:Next wakeup is due at 2011-06-20 13:32:00 (in 100.794154 seconds)

Race condition, which leads to multiple fires of the same job:

guestdev@guestdev-laptop:~/Projet/bac_a_sable/APScheduler-2.0.1$ python test_aps.py 
INFO:apscheduler.threadpool:Started thread pool with 0 core threads and 20 maximum threads
INFO:apscheduler.scheduler:Scheduler started
DEBUG:apscheduler.scheduler:Adding real job with wakeup: True
DEBUG:apscheduler.scheduler:Looking for jobs to run
INFO:apscheduler.scheduler:Added job "job_function (trigger: cron[minute='*/2'], next run at: 2011-06-20 13:32:00)" to job store "default"
DEBUG:apscheduler.scheduler:Next wakeup is due at 2011-06-20 13:32:00 (in 99.725313 seconds)
DEBUG:apscheduler.scheduler:Looking for jobs to run
DEBUG:apscheduler.scheduler:Next wakeup is due at 2011-06-20 13:32:00 (in 99.724925 seconds)

I can't reproduce the bug if I add a job before starting the scheduler.

---

• Bitbucket: https://bitbucket.org/agronholm/apscheduler/issue/13

[agronholm]

Original comment by Alex Grönholm (Bitbucket: agronholm, GitHub: agronholm):

---

This output does not actually demonstrate multiple fires of the same job.
Can you reproduce that? The scheduler waking up more than once is a non-issue.

[agronholm]

Original comment by chemicalstorm (Bitbucket: chemicalstorm, GitHub: chemicalstorm):

---

Nervermind, I am such a dumbass...
I was mislead by the scheduler waking up twice in my example, that I did not see I messed up my cronjob (not the one attached as example in my first post):

#!python
sched.add_cron_job(job_function, **{'minute':'*/2', 'second':'*'})

properly triggers the job every second....

Anyway, thanks for your excellent work on this module! It really awesome and useful.

[agronholm]

Original comment by Alex Grönholm (Bitbucket: agronholm, GitHub: agronholm):

---

Btw, you have a very peculiar way of doing sched.add_cron_job(job_function, minute='/2', second='')