Thank you very very much for the clarification Nathan On Wed, Jan 28, 2015 at 11:01 AM, Nathan Leung <[email protected]> wrote:
> No, the worker is killed and assigned to another worker slot, and that > means any tasks on this worker will also be restarted. It is easier to do > this on the process level than on the task level, and it is more likely to > be effective. > > On Wed, Jan 28, 2015 at 12:21 AM, Sajith <[email protected]> wrote: > >> Hi Nathan, >> >> Thanks for your reply. Could you please ellaborate more on the answer >> please? Is that the failed task will be reinstated and assigned to another >> worker process? >> >> On Tue, Jan 27, 2015 at 7:42 PM, Nathan Leung <[email protected]> wrote: >> >>> The nimbus will detect that the task has timed out and reassign the >>> worker process. >>> >>> On Tue, Jan 27, 2015 at 2:11 AM, Sajith <[email protected]> wrote: >>> >>>> Hi all, >>>> >>>> Can someone please clarify what action does supervisor takes when a >>>> certain tasks fails within a worker process? >>>> >>>> Also, if some other task is emitting tuples to the failed task how does >>>> the emitting task handle the failure of receiving task? >>>> >>>> Is there a way to capture and handler such case? >>>> >>> >>> >> >
