Bumped up number of ackers to 100, which made a *huge* difference--4.3/4.4
million to 6.6 million tuples acked/minute! The capacity of my acker
executors was down around 0.15, so I did not figure I needed to increase
from 10 to 100, but wowsers, that one change made a major impact.

Thanks again to Kobi!

--John

On Mon, Aug 17, 2015 at 12:59 PM, John Yost <[email protected]>
wrote:

> Hi Kobi,
>
> Cool, thanks for getting back to me so quickly!  I did confirm that
> there's one instance of Bolt A (sender, 400 executors) and Bolt B
> (receiver, 100 executors) on each worker (100 workers in topology), so we
> should be good with local shuffling working.
>
> I only have 10 ackers, so I'll bump that up to 100 and see how that works.
>
> Thanks
>
> --John
>
> On Mon, Aug 17, 2015 at 12:26 PM, Kobi Salant <[email protected]>
> wrote:
>
>> Hi John,
>>
>> You should make sure you have at least an instance of each bolt on each
>> worker so local shuffling will work. Also, the number of ackers should be
>> according to the number of workers.
>>
>> Did you check the capacity of the bolts and ackers?
>>
>> Kobi
>>
>> On Mon, Aug 17, 2015 at 7:22 PM, John Yost <[email protected]>
>> wrote:
>>
>>> Hi Everyone,
>>>
>>> I updated my topology to use localOrShuffleGrouping for a Bolt that, for
>>> each incoming tuple, the Bolt generates and emits 15-20 tuples.  My
>>> throughput went from 1 M tuples acked/minute to 4.5 million, which is
>>> great, but I need to get to 7-8 million tuples acked/minute.
>>>
>>> Question--are there any config parameters to use specifically with
>>> localOrShuffleGrouping?  Please confirm, thanks!
>>>
>>> --John
>>>
>>
>>
>> This message may contain confidential and/or privileged information.
>> If you are not the addressee or authorized to receive this on behalf of
>> the addressee you must not use, copy, disclose or take action based on this
>> message or any information herein.
>> If you have received this message in error, please advise the sender
>> immediately by reply email and delete this message. Thank you.
>>
>
>

Reply via email to