If you want to send tuples to local worker process from one task to other,
you can use `localOrShuffleGrouping` between spout and bolt. But I do not
think we have inbuilt grouping which sends it to any of the workers on the
same supervisor.
For ex:
TopologyBuilder builder = new TopologyBuilder();
builder.setSpout("spout", new RandomSentenceSpout(), 5);
builder.setBolt("split", new SplitSentence(),
8). localOrShuffleGrouping("spout");
Thanks,
Satish.
On Fri, Jun 17, 2016 at 2:28 PM, hjh <[email protected]> wrote:
> Thank you for your kind reply! Well, this is a special using case, we need
> to make the spout or bolt tasks to transmit data to the tasks, that run on
> the same supervisor or machine, of the next bolt. For example, Spout A,
> task 1 runs on Supervisor 1, Bolt A is connected to Spout A and it has two
> tasks where task 1 run on Supervisor 1 and task 2 run on Supervisor 2, and
> we need to route data from Spout A task 1 to Bolt A task 1 not to task 2.
> And it is better to not introduce the third party as a central coordinator.
> Thank you very much!
>
> By the way, to do so, it is also required that some of the scheduling
> information to be shared globally in the custom scheduler. And I am not
> sure if this is possible to do that. Thank you!
>
> Best regards!
>
>
> On 06/17/2016 06:10 AM, Satish Duggana wrote:
>
> Hi,
> Why do you need supervisor-id in a bolt/spout task? What are you using
> for?
>
> Thanks,
> Satish.
>
> On Fri, Jun 17, 2016 at 2:13 AM, applyhhj <[email protected]> wrote:
>
>> Hello，everyone！
>> Is there anyway to get the id of the supervisor that the Bolt or the
>> Spout is running on? And the also the id of the supervisor that the
>> consumer tasks are running on? Thank you in advance!!
>>
>> Best regards!
>>
>> 2016-06-16
>> ------------------------------
>> applyhhj
>>
>
>
>
