How to: Setup the multi-clustering on the Nodegrid OS

How to: Setup the multi-clustering on the Nodegrid OS

This article deals with the description and the configuration of the multi-clustering feature on the Nodegrid OS.

Description:
  1. Multi-Clustering allows a centralized view and access across various clusters, therefore across various Datacenters and Sites. 
  2. Its design allows a coordinator of a cluster (possibly having only itself as a node) to "join" another cluster: it will be capable of seeing the remote coordinator, its peers and corresponding devices. It will be able to see all devices of the remote cluster, regardless of the topology of the remote cluster, either mesh or star. 
  3. This is directional: when a coordinator of cluster A joins cluster B, the coordinator of cluster A can see the peers and devices of cluster B, but the nodes of cluster B, including its coordinator, cannot see the nodes of cluster A. If you want the other direction, the coordinator of cluster B needs to join cluster A.
  4. It allows the customer to have N clusters, mesh and/or star, and have a "super coordinator" to join each one of the N clusters, thus being capable of seeing all nodes and devices. If there is any concern on this being a single point of failure, just add another "super coordinator" with the same configuration.


  1. Note: The "super coordinator" concept is not required. If one has two clusters, one on each datacenter, the coordinator of one joins the other and vice-versa, so both coordinators have visibility of both clusters. This design aims to reduce the number of live connections and data that needs to be propagated, in favor of the scalability of the solution.
  2. Available from Nodegrid OS version 5.8.0 and above. 
Configuration:
  1. In System -> License: A Clustering license is required.
  2. Configure a Cluster for the SuperCoordinator:  It can be composed of one node only. Choose the Star mode: 


  1. Configure each Cluster as usual: the idea is that they match the urbanization (one per DC, one per industrial site, …). They can be either in Mesh or Star mode: 

  1. The SuperCoordinator joins the various Clusters: In Cluster -> Clusters: Selection and Join: 
Then Add each Cluster, entering the Cluster Name, the Cluster coordinator’s IP address, and the Cluster’s PSK. 

Visualize the result in Cluster -> Clusters: the various clusters appear, with the IP address of the respective coordinators and with the status “On-line”: 


Testing:
You can visualize the results: 
  1. On the map: In Access -> Map:

  1. On the Centralized Access: in Access -> Table: 


Note: the features associated with the Centralized Access of the usual Clusters remain applicable: 
  1. In Cluster :: Management : mass software upgrades. 
  2. In Systems :: Central Management:  execution of Ansible Playbooks from the WebUI.