If you're going to upgrade, I would recommend jumping to the latest
version, 1.9.2 as of today. We ran into some issues in 1.8 with this
feature that was fixed in 1.9.x. We're running 1.9.2 now with this feature
now in production.

Regards

Dano

On Tue, Jul 9, 2019, 6:58 AM <josef.zahn...@swisscom.com> wrote:

> The feature requires NiFi > 1.8.x… Pierre describes it very well in his
> blog :
> https://pierrevillard.com/2018/10/29/nifi-1-8-revolutionizing-the-list-fetch-pattern-and-more/
>
>
>
>
>
> *From: *James McMahon <jsmcmah...@gmail.com>
> *Reply-To: *"users@nifi.apache.org" <users@nifi.apache.org>
> *Date: *Tuesday, 9 July 2019 at 14:46
> *To: *"users@nifi.apache.org" <users@nifi.apache.org>
> *Subject: *Re: DistributeLoad across a NiFi cluster
>
>
>
> Andrew, when I right click on the connection between the two I do not see
> a cluster distribution strategy in the queue connection. I am running
> 1.7.1.g. Am I overlooking something?
>
>
>
> On Tue, Jul 2, 2019 at 12:29 PM Andrew Grande <apere...@gmail.com> wrote:
>
> Jim,
>
>
>
> There's a better solution in NiFi. Right click on the connection between
> ListFile and FetchFile and select a cluster distribution strategy in
> options. That's it :)
>
>
>
> Andrew
>
>
>
> On Tue, Jul 2, 2019, 7:37 AM James McMahon <jsmcmah...@gmail.com> wrote:
>
> We would like to employ a DistributeLoad processor, restricted to run on
> the primary node of our cluster. Is there a recommended approach employed
> to efficiently distribute across nodes in the cluster?
>
>
>
> As I understand it, and using a FetchFile running in "all nodes" as the
> first processor following the DistributeLoad, I can have it distribute by
> round robin, next available, or load distribution service.  Can anyone
> provide a link to an example that employs the load distribution service? Is
> that the recommended distribution approach when running in clustered mode?
>
>
>
> I am interested in maintaining load balance across my cluster nodes when
> running at high flowfile volumes. Flow files will vary greatly in contents,
> so I'd like to design with an approach that helps me balance processing
> distribution.
>
>
>
> Thanks very much in advance. -Jim
>
>

Reply via email to