@Pierre Villard We are currently using Nifi 1.7 over 10 nodes where the nodes are distributed across two data centers. The design uses an external Zookeeper. However, we are facing a problem where the Primary Co-ordinator being elected at random by the Zookeeper across both the data centers, and electing particular servers as Primary having higher I/O and thus increasing the CPU while doing the List function. Is there a way we can restrict the election of the Primary to particular servers by the external Zookeeper ? I am looking for a solution in Nifi 1.7. Any suggestions will be very helpful.
I know this doesn't help you in NiFi 1.7.0, but if you are able to upgrade to 1.8.0, there is some new Load Balancing functionality in the latest version that might be of help to you. There is a good blog on this (posted on 10/29/2018 with the title "Load Balancing Data Across a NiFi Cluster"):
and here is the link to the related documentation: