I know that my problem is in the facebook bot. the latency there with the http requests. Wonder how I could improve it to have less failed bolts. I added "Deduplicator" bolt that avoid duplications of same messages but still I need to eliminate the failed ones as much as possible
2015-03-12 14:47 GMT+02:00 Nathan Leung <[email protected]>: > Tuples trees are tracked by storm in their entirety as long as you anchor > new tuples properly. This is the setting you are looking for. > On Mar 12, 2015 8:41 AM, "vahid islami" <[email protected]> wrote: > >> thanks , but I need limit message on whole storm not only spout. max >> spout only limit spout and when first bolt receive message form spout and >> process that , this bolt ack back to spout and message aggregate in middle >> storm. >> >> On Thu, Mar 12, 2015 at 3:48 PM, <[email protected]> wrote: >> >>> You can use max spout spending to limit your spout's tuples >>> >>> Sent from my iPhone >>> >>> > On Mar 12, 2015, at 6:41 AM, vahid islami <[email protected]> >>> wrote: >>> > >>> > Hi , >>> > I work with storm and ActiveMQ. I need to limit the number message in >>> storm. for example if spout send 40 message to next bolts and until the >>> message not process, spout not send message. >>> > I can handle if final bolt ack back to spout or I notify ack event in >>> bolts. >>> > how implement this ? >>> > thank you >>> >> >>
