I guess using jsonsole or using some profiling mechanism try to identify which part of processing in the worker is consuming more memory ...
On Tue, Aug 5, 2014 at 11:32 AM, Vladi Feigin <[email protected]> wrote: > These are lot of logs and they are relatively large . Could you point me > to the specific details to look for..? > Vladi > > > On Tue, Aug 5, 2014 at 8:28 AM, padma priya chitturi < > [email protected]> wrote: > >> Can you share with us the worker logs. >> >> --Padma Ch >> >> >> On Tue, Aug 5, 2014 at 12:21 AM, Vladi Feigin <[email protected]> wrote: >> >>> Hi, >>> >>> Recently we observed very strange Storm behavior : >>> We had only one worker process, among dozens with very high memory usage >>> (a few gigs). We understand that this is not Java objects , since our heap >>> size is only 1G. We don't use field grouping in this topology so this is >>> not a hot spot. >>> Does someone had problems like that? How can we investigate the cause? >>> >>> We use Storm version 0.82 >>> Thank you in advance, >>> Vladi >>> >>> >> >> >
