That is correct... Flow files don't move between nodes unless the flow specifically does so using a remote process group pointing back to itself.
An example of when you would do this would be something like the List + Fetch pattern. This article shows some examples: https://community.hortonworks.com/articles/16120/how-do-i-distribute-data-across-a-nifi-cluster.html -Bryan On Mon, Jan 8, 2018 at 6:01 PM, FoodEssentials <[email protected]> wrote: > Holy crap. So does that mean if a source processor runs on node 1, all > subsequent flowfiles from that source will also run on node 1? This makes a > lot more sense now. > If that’s true then a lot of our problems with NiFi are due to user error. > > Thanks > > > On Jan 8, 2018, at 4:51 PM, Bryan Bende <[email protected]> wrote: > > Primary-node-only is really just a way to have source processors run > on only one node in the cluster. > > All downstream processors should be running on all nodes so that it > doesn't matter if the source processors fail over to a different > primary node. > >
