Dear Alexander, Thank you very much for firing up some neurons ... very much helpful :)
Regards, Sreeram On Fri, Mar 18, 2016 at 8:43 PM, Alexander T <mittspamko...@gmail.com> wrote: > Hi Sreeram, > > Some ideas: One way to tolerate replays would be to use some type of CRDT > counter (https://en.wikipedia.org/wiki/Conflict-free_replicated_data_type). > Poor mans CRDT could be constructed if you have a strict ordering of your > messages through a monotonically increasing counter. Save the message id > when you update the count and only update if your message id is larger than > the last saved. This could also be done sloppily by using timestamps. If > you don't want to store the extra state in HBase you could utilize a > stateful bolt to store it. > > Best regards, > > Alexander > > On Fri, Mar 18, 2016 at 10:37 AM, Sreeram <sreera...@gmail.com> wrote: > >> Hi, >> >> I am looking for suggestions from community on implementing HBase atomic >> increment in a idempotent manner from inside a storm bolt. >> >> Replay of the storm bolt results in a double increment. >> >> Any suggestion on the right approach to be taken is welcome. >> >> Thank you. >> >> Regards, >> Sreeram >> > >