I've prepared a minimal patch to remove this behavior from the scheduler: https://reviews.apache.org/r/32840/
-=Bill On Fri, Apr 3, 2015 at 3:50 PM, Bill Farner <wfar...@apache.org> wrote: > The bright side is that you (the cluster administrator) can buffer for > this in only the global task timeout setting, and it doesn't have to leak > into everyone's update configurations. > > -=Bill > > On Fri, Apr 3, 2015 at 1:06 PM, Steve Niemitz <st...@tellapart.com> wrote: > >> oh hm, I thought it started as soon as the task was launched, if it >> already >> starts the timer at RUNNING never mind then :) >> >> On Fri, Apr 3, 2015 at 12:18 PM, Maxim Khutornenko <ma...@apache.org> >> wrote: >> >> > >Does this also include making the timer for "initial_interval_secs" >> start >> > >when the task enters RUNNING? >> > >> > Did you rather mean STARTING as in AURORA-894? The >> > "initial_interval_sec" currently starts in RUNNING and AURORA-894 >> > suggests moving it into STARTING. >> > >> > On Fri, Apr 3, 2015 at 12:02 PM, Steve Niemitz <st...@tellapart.com> >> > wrote: >> > > +100. This is particularly annoying with docker tasks, as the task >> > doesn't >> > > enter running until the pull completes. >> > > >> > > Does this also include making the timer for "initial_interval_secs" >> start >> > > when the task enters RUNNING? >> > > >> > > On Fri, Apr 3, 2015 at 11:57 AM, Joseph Smith <yasumo...@gmail.com> >> > wrote: >> > > >> > >> Also a +1, users can’t do anything to modify this other than just >> > increase >> > >> this threshold in case we don’t do the right thing. >> > >> >> > >> > On Apr 2, 2015, at 4:58 PM, Joshua Cohen <jco...@twopensource.com> >> > >> wrote: >> > >> > >> > >> > +1 >> > >> > >> > >> > On Thu, Apr 2, 2015 at 2:34 PM, Zameer Manji <zma...@apache.org> >> > wrote: >> > >> > >> > >> >> +1 >> > >> >> >> > >> >> I think we should be very careful in the future to ensure user >> > provided >> > >> >> timeouts do not cover state transitions outside of their control. >> > >> >> >> > >> >> On Thu, Apr 2, 2015 at 2:29 PM, David McLaughlin < >> > da...@dmclaughlin.com >> > >> > >> > >> >> wrote: >> > >> >> >> > >> >>> +1 to the proposal. >> > >> >>> >> > >> >>> On Thu, Apr 2, 2015 at 2:09 PM, Maxim Khutornenko < >> ma...@apache.org >> > > >> > >> >>> wrote: >> > >> >>> >> > >> >>>> We are proposing to drop the UpdateConfig restart_threshold >> setting >> > >> >>>> due to not delivering much end user value. Please, take a >> moment to >> > >> >>>> read through https://issues.apache.org/jira/browse/AURORA-1240 >> and >> > >> >>>> reply back if you have any doubts. >> > >> >>>> >> > >> >>>> Thanks, >> > >> >>>> Maxim >> > >> >>>> >> > >> >>> >> > >> >>> -- >> > >> >>> Zameer Manji >> > >> >>> >> > >> >>> >> > >> >> >> > >> >> > >> >> > >> > >
