Hey, I recommend 1.0 for now and see if that solves the no ack issue. I recommend max.spout.pending at 10K to hopefully ensure you avoid tuple failures and then move up from there if you see back pressure. Back pressure you'll see in the form of Bolt capacities way less than 0.5.
--John On Sat, Jan 16, 2016 at 12:29 PM, researcher cs <[email protected]> wrote: > @john thanks for your time and help , i need to know is value 1.0 is the > best value should i use or can i increase it ? > max.spout.pounding = 50000 or 50K so what are other properties should i > change from default value in storm.yaml like > topology.acker.executors > topology.executor.receive.buffer.size 1024 > topology.executor.send.buffer.size 1024 > > i have one worker - one supervisor -- > > > > On Sat, Jan 16, 2016 at 1:39 PM, John Yost <[email protected]> wrote: > >> Hi, >> >> Argh, I apologize, just getting back to this. The storm default >> topology.stats.sample.rate is 0.05, I set it to 1.0 for mine, and that's >> when I started seeing numbers for executed/emitted/acked. I think the >> sample rate needs to be higher for lower throughput installations like >> mine, and maybe your's as well. >> >> Again, sorry for the delay--hope this works for you also! >> >> --John >> >> >> On Wed, Jan 13, 2016 at 9:53 AM, Stephen Powis <[email protected]> >> wrote: >> >>> >>> http://stackoverflow.com/questions/26276096/performace-impact-of-using-setstatssamplerate-topology-stats-sample-rate >>> >>> On Wed, Jan 13, 2016 at 9:51 PM, researcher cs < >>> [email protected]> wrote: >>> >>>> what is this properties used for topology.stats.sample.rate ? >>>> >>> >>> >> >
