ZooKeeper acts as a coordination service that provides flexible and robust synchronization within distributed systems. Apache ZooKeeper is already packaged with the ZENworks 2020 build. When you upgrade or install the ZENworks 2020 or a later release version on your first Primary Server, the ZooKeeper role is automatically enabled on this server.
In ZENworks, ZooKeeper runs as a thread within the ZENworks Server services and is used by various components including the Vertica database. Therefore you need to ensure that ZooKeeper is up and running at all times to enable proper functioning of the various ZENworks components. To view the status of ZooKeeper, navigate to the Diagnostics page in ZCC. For more information, see ZENworks Diagnostics and Probe Guide.
When ZooKeeper is enabled on the first Primary Server, then the default port 6789 is automatically opened. If at any point in time the ZooKeeper is down, then try restarting the ZENserver services. However, if the service is still down, then ensure that the firewall allows client connections from other Primary Servers to the ZooKeeper services on port 6789. If the Primary Servers in your zone are unable to access the ZooKeeper service due to firewall restrictions, then to open the ports, you can run the following Configure action on the server in which ZooKeeper is enabled
novell-zenworks-configure -c ClusterFirewallConfigureAction -Doperation=add -Dservice=zookeeper
However, if the Primary Server that is within the DMZ is unable to access the ZooKeeper service within the corporate network, then you need to manually open the port 6789 in the corporate firewall.
For more information on the ZooKeeper ports, see ZENworks TCP and UDP Ports.
You can expand the ZooKeeper cluster as per your scalability requirements and to provide fault tolerance. To expand the ZooKeeper cluster and to maintain the cluster you need to execute a set of ZMAN commands.
Apache ZooKeeper is already packaged with the ZENworks 2020 build. When the build is deployed on the first Primary Server, ZooKeeper is automatically enabled in the zone. However, you can expand the ZooKeeper cluster as per your scalability requirements and to provide fault tolerance. It is recommended to have an odd number of servers in the cluster. A three node cluster still continues to function if one server goes down since the majority servers are still up. A four node cluster also can only survive the loss of a single node since if two nodes go down, the cluster loses its majority. Therefore, no additional fault tolerance that can be achieved by configuring the cluster with an even number of servers.