Nicolas, The target URL is used when first making the connection. When you start the RPG it will connect to the target URL to learn about all the nodes in the cluster, after that it is talking directly to all nodes and handling failover.
So the scenario where you have a problem is if your RPG is restarted and the target URL happens to be down at that moment. To address that there has been some work done to eventually allow entering multiple URLs: https://github.com/apache/nifi/pull/1208 Thanks, Bryan On Mon, Nov 28, 2016 at 9:00 AM, Provenzano Nicolas < [email protected]> wrote: > Hi all, > > > > As described in many topics, the processing of collected files can be > distributed over a Nifi cluster by using a remote process group. > > > > For example, we can use a listSFTP processor, connect it to a port that is > the input of a remote process group. > > > > However, when configuring the URL of the RPG, we must set it with an IP > address that can be any node of the cluster. However, if the node goes > down, the flow is down… > > > > Is there any way by setting the URL of the RPG in order to use the elected > primary node ? The input of the RPG would then be dynamically changed > according to the elected node. > > > > If it’s not possible, how can we set it to avoid the problem I described ? > > > > Thanks in advance > > > > BR > > > > Nicolas > > > > — > Gfi Informatique Telecom > *Nicolas Provenzano* > > BU Telecom / Customer Experience > > *[email protected] <[email protected]>* > — > 2 rue Andras Beck – 92360 Meudon La Forêt > Tél. : +33 (0)1 46 83 35 06 > Mob. : +33 (0)6 82 07 19 21 > www.gfi.world > — > [image: signature-img] > > >
