[
https://issues.apache.org/jira/browse/STORM-2755?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16179307#comment-16179307
]
Stig Rohde Døssing commented on STORM-2755:
-------------------------------------------
Hi Oussama,
Yes, this is what stream groupings are for. Please see
https://storm.apache.org/releases/2.0.0-SNAPSHOT/Concepts.html under the
"Stream Groupings" header. With an appropriate grouping you can make sure that
all related data goes to the same bolt instance. I would recommend you start by
looking at the fields grouping. If you'd like a code example, the
WordCountTopology uses a field grouping here
https://github.com/apache/storm/blob/master/examples/storm-starter/src/jvm/org/apache/storm/starter/WordCountTopology.java#L88
> Sharing data between Bolts
> --------------------------
>
> Key: STORM-2755
> URL: https://issues.apache.org/jira/browse/STORM-2755
> Project: Apache Storm
> Issue Type: Question
> Reporter: Oussama BEN LAMINE
>
> Hello,
> i am facing a problem and your help will be really appreciated.
> so, i am using storm 1.0.1, i prepared a topology that transfers data throw a
> preparation bolt then to a python prediction bolt.
> in prediction bolt i am collecting data in real time and creating group of
> data then sending them to prediction function.
> this topology works perfectly in one machine where the machine can process
> all the data, but not in distributed cluster as each machine work separately
> and some data needed for grouping can be processed by different machine.
> Is there a way that bolts communicate with each other and so we can group
> data even if they are on different machines?
> can you please advice and thank you
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
