Thanks Bill a lot for the quick reply! I was trying to find the documentation about all of this this but failed. Regarding to the dead tasks, i was encountering one issue in rescheduling a cron job. i can see the job became "adhoc" status on GUI. But i can't reschedule it (reported "Job xxx already exists") and also can't deschedule it (reported "Job xxx is not scheduled with cron"). Any thoughts ?
On Sat, Jun 11, 2016 at 10:20 PM, Bill Farner <[email protected]> wrote: > May i ask if there is a way to completely remove cron/service job from >> Aurora ? >> For cron job, when i deschedule them, they become adhoc jobs. >> For service job, what i can do is killall them. >> Still on the Aurora UI, i can still see all of the tasks. > > > Dead tasks (the remnants of a killed job) are reclaimed eventually so that > they may still be observed, by default in at most 2 days [1] or if a job's > history becomes lengthy [2]. There is no mechanism to forcibly delete > them. In practice, i've found this to do more good than harm. > > Another question is for service job, how can i make Aurora reschedule it >> immediately when failure. From GUI, i saw "throttling" status reporting. > > > Aurora will back off from rescheduling a service task when it runs but > exits quickly [3]. This is to prevent starvation of the scheduler in > large/busy clusters. The best course of action is to stabilize the job so > that it remains up longer. > > [1] > https://github.com/apache/aurora/blob/master/src/main/java/org/apache/aurora/scheduler/pruning/PruningModule.java#L41-L44 > [2] > https://github.com/apache/aurora/blob/master/src/main/java/org/apache/aurora/scheduler/pruning/PruningModule.java#L46-L48 > [3] > https://github.com/apache/aurora/blob/master/src/main/java/org/apache/aurora/scheduler/scheduling/SchedulingModule.java#L43-L56 > > On Sat, Jun 11, 2016 at 1:19 AM, Ziliang Chen <[email protected]> > wrote: > >> Hi, >> >> May i ask if there is a way to completely remove cron/service job from >> Aurora ? >> For cron job, when i deschedule them, they become adhoc jobs. >> For service job, what i can do is killall them. >> Still on the Aurora UI, i can still see all of the tasks. >> >> Another question is for service job, how can i make Aurora reschedule it >> immediately when failure. From GUI, i saw "throttling" status reporting. >> >> Thank you ! >> >> -- >> Regards, Zi-Liang >> >> Mail:[email protected] >> > > -- Regards, Zi-Liang Mail:[email protected]
