Skip to content

Instantly share code, notes, and snippets.

@mplewis
Last active December 15, 2023 13:09
Show Gist options
  • Star 34 You must be signed in to star a gist
  • Fork 8 You must be signed in to fork a gist
  • Save mplewis/8483f1c24f2d6259aef6 to your computer and use it in GitHub Desktop.
Save mplewis/8483f1c24f2d6259aef6 to your computer and use it in GitHub Desktop.
An implementation of Scheduler that catches jobs that fail. For use with https://github.com/dbader/schedule
import logging
from traceback import format_exc
import datetime
from schedule import Scheduler
logger = logging.getLogger('schedule')
class SafeScheduler(Scheduler):
"""
An implementation of Scheduler that catches jobs that fail, logs their
exception tracebacks as errors, optionally reschedules the jobs for their
next run time, and keeps going.
Use this to run jobs that may or may not crash without worrying about
whether other jobs will run or if they'll crash the entire script.
"""
def __init__(self, reschedule_on_failure=True):
"""
If reschedule_on_failure is True, jobs will be rescheduled for their
next run as if they had completed successfully. If False, they'll run
on the next run_pending() tick.
"""
self.reschedule_on_failure = reschedule_on_failure
super().__init__()
def _run_job(self, job):
try:
super()._run_job(job)
except Exception:
logger.error(format_exc())
job.last_run = datetime.datetime.now()
job._schedule_next_run()
#!/usr/bin/env python3
import time
from safe_schedule import SafeScheduler
def good_task_1():
print('Good Task 1')
def good_task_2():
print('Good Task 2')
def good_task_3():
print('Good Task 3')
def bad_task_1():
print('Bad Task 1')
print(1/0)
def bad_task_2():
print('Bad Task 2')
raise Exception('Something went wrong!')
scheduler = SafeScheduler()
scheduler.every(3).seconds.do(good_task_1)
scheduler.every(5).seconds.do(bad_task_1)
scheduler.every(7).seconds.do(good_task_2)
scheduler.every(8).seconds.do(bad_task_2)
scheduler.every(12).seconds.do(good_task_3)
while True:
scheduler.run_pending()
time.sleep(1)
@eddified
Copy link

eddified commented Mar 18, 2019

Hey @mplewis
im new to python and im wonder about how reschedule_on_failure=True work?
seems has no effect to anything
pls help me understand of this

Thanks <3

You're right, I don't think reschedule_on_failure does anything. I checked the parent class (version 0.6.0) and it doesn't reference that variable at all.

@fimasini
Copy link

fimasini commented Jan 5, 2021

Hi,
I modified your extension in a way that the job can be rescheduled in minutes o seconds after a failure, or can be even canceled on failure.

Thanks for the work.

Bye

class SafeScheduler(Scheduler):
    """
    An implementation of Scheduler that catches jobs that fail, logs their
    exception tracebacks as errors, optionally reschedules the jobs for their
    next run time, and keeps going.

    Use this to run jobs that may or may not crash without worrying about
    whether other jobs will run or if they'll crash the entire script.
    """

    def __init__(self, reschedule_on_failure=True, minutes_after_failure=0, seconds_after_failure=0):
        """
        If reschedule_on_failure is True, jobs will be rescheduled for their
        next run as if they had completed successfully. If False, they'll run
        on the next run_pending() tick.
        """
        self.reschedule_on_failure = reschedule_on_failure
        self.minutes_after_failure = minutes_after_failure
        self.seconds_after_failure = seconds_after_failure
        super().__init__()

    def _run_job(self, job):
        try:
            super()._run_job(job)
        except Exception:
            logger.error(format_exc())
            if(self.reschedule_on_failure):
                if(self.minutes_after_failure!=0 or self.seconds_after_failure!=0):
                    logger.warn("Rescheduled in %s minutes and %s seconds." % (self.minutes_after_failure, self.seconds_after_failure))
                    job.last_run = None
                    job.next_run = datetime.now() + timedelta(minutes=self.minutes_after_failure, seconds=self.seconds_after_failure)
                else:
                    logger.warn("Rescheduled.")
                    job.last_run = datetime.now()
                    job._schedule_next_run()
            else:
                logger.warn("Job canceled.")
                self.cancel_job(job)

@claudineien
Copy link

Hi @fimasini , @mplewis

I got the following error " AttributeError: module 'datetime' has no attribute 'now' ", so I changed datetime.now() to datetime.datetime.now().

@fimasini Thanks for the contributed
@mplewis Thanks for awesome initiative

class SafeScheduler(Scheduler):
    """
    An implementation of Scheduler that catches jobs that fail, logs their
    exception tracebacks as errors, optionally reschedules the jobs for their
    next run time, and keeps going.

    Use this to run jobs that may or may not crash without worrying about
    whether other jobs will run or if they'll crash the entire script.
    """

    def __init__(self, reschedule_on_failure=True, minutes_after_failure=0, seconds_after_failure=0):
        """
        If reschedule_on_failure is True, jobs will be rescheduled for their
        next run as if they had completed successfully. If False, they'll run
        on the next run_pending() tick.
        """
        self.reschedule_on_failure = reschedule_on_failure
        self.minutes_after_failure = minutes_after_failure
        self.seconds_after_failure = seconds_after_failure
        super().__init__()

    def _run_job(self, job):
        try:
            super()._run_job(job)
        except Exception:
            logger.error(format_exc())
            if(self.reschedule_on_failure):
                if(self.minutes_after_failure!=0 or self.seconds_after_failure!=0):
                    logger.warn("Rescheduled in %s minutes and %s seconds." % (self.minutes_after_failure, self.seconds_after_failure))
                    job.last_run = None
                    job.next_run = datetime.datetime.now() + timedelta(minutes=self.minutes_after_failure, seconds=self.seconds_after_failure)
                else:
                    logger.warn("Rescheduled.")
                    job.last_run = datetime.datetime.now()
                    job._schedule_next_run()
            else:
                logger.warn("Job canceled.")
                self.cancel_job(job)

@dumaaan
Copy link

dumaaan commented Sep 27, 2021

Hello! I have noticed that if my job keeps failing too many times, the entire process gets killed because I get queue.Full error. Any way to clear that up?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment