You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@nifi.apache.org by "Raffaele Saggino (Jira)" <ji...@apache.org> on 2021/03/15 10:16:00 UTC

[jira] [Commented] (NIFI-6972) Stateless NiFi should support parallelism

    [ https://issues.apache.org/jira/browse/NIFI-6972?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17301566#comment-17301566 ] 

Raffaele Saggino commented on NIFI-6972:
----------------------------------------

> In other words, it runs only one flow file at a time. Is there any reason it was implemented that way?
This ensures full ordering on the pipeline and should also avoid duplicates.

Not sure if this was an original requirement.

Raffaele

> Stateless NiFi should support parallelism
> -----------------------------------------
>
>                 Key: NIFI-6972
>                 URL: https://issues.apache.org/jira/browse/NIFI-6972
>             Project: Apache NiFi
>          Issue Type: Improvement
>          Components: NiFi Stateless
>    Affects Versions: 1.10.0
>            Reporter: Adam Trump
>            Priority: Major
>              Labels: features
>
> Today, when running NiFi in stateless mode, it will run all flow files serially - meaning that only after a flow file finish running the whole flow, it will re run the flow for the next flow file.
> In other words, it runs only one flow file at a time. Is there any reason it was implemented that way?
> Instead, I think that there should be a new parameter when running the NiFi, which would be the maximum parallelism. In my opinion, it could be implemented quite easily:
> Under StatelessFlow.java, in the run(final Queue<InMemoryFlowFile> output) method, in line 322, it should create a new thread each iteration, with the maximum being the given parameter.
> This will allow the flow to run multiple flow file in parallel, and should be simple enough without breaking anything else...



--
This message was sent by Atlassian Jira
(v8.3.4#803005)