rkiddy / apptimer Goto Github PK

In my logs I am seeing that some tasks run over a dozen times a day. When I restart the app, this does not occur.

I am observing that if a task instance ends in an error, then there will be a burst of instances. A subsequent successful run will not stop the rush of task instances being run.

The scheduler should not be multiply enqueueing, because I check for isQueued and isRunning at the start of the schedule check.

For example, I see these instances.

Task                    Queued                  Started                 Ended                 Runtime Result
DailyYubPopReader 	2021-04-05 00:12:49 	2021-04-05 00:14:06 	2021-04-05 00:14:27 	20 	0 	
DailyYubPopReader 	2021-04-04 12:12:01 	2021-04-04 12:12:01 	2021-04-04 12:12:21 	20 	0 	
DailyYubPopReader 	2021-04-04 12:09:53 	2021-04-04 12:09:53 	2021-04-04 12:10:13 	20 	0 	
DailyYubPopReader 	2021-04-04 12:07:48 	2021-04-04 12:07:48 	2021-04-04 12:08:08 	20 	0 	
DailyYubPopReader 	2021-04-04 12:05:38 	2021-04-04 12:05:38 	2021-04-04 12:05:58 	20 	0 	
DailyYubPopReader 	2021-04-04 12:03:29 	2021-04-04 12:03:29 	2021-04-04 12:03:49 	19 	0 	
DailyYubPopReader 	2021-04-04 12:01:25 	2021-04-04 12:01:25 	2021-04-04 12:01:45 	19 	0 	
DailyYubPopReader 	2021-04-04 12:00:25 	2021-04-04 12:00:25 	2021-04-04 12:00:45 	20 	0 	
DailyYubPopReader 	2021-04-04 11:59:13 	2021-04-04 11:59:13 	2021-04-04 11:59:33 	20 	0 	
DailyYubPopReader 	2021-04-04 11:57:48 	2021-04-04 11:57:48 	2021-04-04 11:58:21 	32 	2 	no such element
DailyYubPopReader 	2021-04-04 11:56:35 	2021-04-04 11:56:35 	2021-04-04 11:56:55 	20 	0 	
DailyYubPopReader 	2021-04-04 11:55:18 	2021-04-04 11:55:18 	2021-04-04 11:55:38 	19 	0 	
DailyYubPopReader 	2021-04-04 11:54:15 	2021-04-04 11:54:15 	2021-04-04 11:54:36 	20 	0 	
DailyYubPopReader 	2021-04-04 11:53:15 	2021-04-04 11:53:15 	2021-04-04 11:53:36 	21 	0 	
DailyYubPopReader 	2021-04-04 11:52:28 	2021-04-04 11:52:28 	2021-04-04 11:52:49 	20 	0 	
DailyYubPopReader 	2021-04-04 11:00:44 	2021-04-04 11:46:33 	2021-04-04 12:12:21 	1547 	0 	
DailyYubPopReader 	2021-04-03 22:58:40 	2021-04-03 23:00:05 	2021-04-03 23:00:26 	21 	0 	
DailyYubPopReader 	2021-04-03 22:55:00 	2021-04-03 22:57:10 	2021-04-03 22:58:07 	57 	2 	no such element
DailyYubPopReader 	2021-04-03 10:51:05 	2021-04-03 10:53:45 	2021-04-03 10:54:12 	26 	0 	
DailyYubPopReader 	2021-04-02 22:49:55 	2021-04-02 22:50:18 	2021-04-02 22:50:35 	16 	0 	

Right now, the scheduling just says "what is the time elapsed since the last successful run?" But this means that if a job fails, it will try to run again with every scheduling interval. Is this what is desired?

Should there also be a "failure schedule" defined? This would be a separate column in the app_tasks table. It would, I think, think take values like "TRY_HOURLY", "TRY_HOURS_2, or "TRY_MINUTES_15".

For example, something is going wrong in my pdf processing on dailyjail and the jobs are repeating over and over again.

If a job repeats more than some number of times, it should deactivate itself and send me a warning.

Well, at least one is not required to run it and it does rely on the Invitations framework to run but it will not break the build if the Invitations framework is not available.

I have seen that when a long task is running, the RelationChecker task gets queued more than once. This means that, when it comes up, several of them run consecutively, which is not the intention.

Need to confirm this supposition.

See https://opencalaccess.org/img/Screen%20Shot%202021-03-26%20at%2012.27.20%20PM.png

While something is running or queued, it would be good to see how long things are taking, so that we can tell when something is broken.

I cannot do this now. I will need to be able to either identify the thread running the task and kill it, or I need the Future object that the submit() method of the Executor provided, which I can use to kill the task.

I need the ability to track the Future objects for other reasons, so that will probably be best.

Perhaps, but my scheduler does something different.

The cron4j framework allows one to schedule jobs with cron-like schedule strings, such as "16 14 * * *", which will run something at 4:16 pm every day. And there are lots of options for smart ways to specify various scheduling concepts.

But it does say "run my thing at this or that time or times" and what I wanted was "run my thing after this long a duration".

Is it better to say "I want this to run at 10 am and 10 pm", or it is better to say "I want this to run every 12 hours"? I am not sure.