Synchronization across cluster nodes
All configurations on a NetScaler cluster are performed on the cluster IP address, which is the management address of the cluster. This cluster IP address is owned by a cluster node that is referred to as the cluster configuration coordinator as shown in the following figure:
The configurations that are available on the configuration coordinator are automatically propagated to the other cluster nodes and therefore all cluster nodes have the same configurations.
- NetScaler appliance allows only a few configurations to be performed on individual cluster nodes through their NSIP address. In these cases, you must ensure configuration consistency manually across all nodes in the cluster. These configurations are not propagated across the other cluster nodes. For more information on operations supported on each cluster nodes, see “Operations Supported on Individual Cluster Nodes”.
- The following commands when executed on the cluster IP address are not propagated to other cluster nodes:
- shutdown. Shuts down only the configuration coordinator.
- reboot. Reboots only the configuration coordinator.
- rm cluster instance. Removes the cluster instance from the node that you are executing the command on.
- For a command to propagate to other cluster nodes:
- The quorum must be configured on cluster instance.
- The majority of the cluster quorum with (n/2 + 1) of the cluster nodes must be active for the cluster to be operational.
- A cluster can run with a minimum number of nodes when majority rule (n/2 + 1) is relaxed.
When a node is added to a cluster, the configurations and the files (SSL certificates, licenses, DNS, and so on) that are available on the cluster configuration coordinator are synchronized to the newly added cluster node. When an existing cluster node, that was intentionally disabled or that had failed, is once again added, the cluster compares the configurations available on the node with the configurations available on the configuration coordinator. If there is a mismatch in configurations, the node is synchronized by using one of the following:
- Full synchronization. If the difference between configurations exceeds 255 commands, all the configurations of the configuration coordinator are applied to the node that is rejoining the cluster. The node remains operationally unavailable for the duration of the synchronization.
- Incremental Synchronization. If the difference between configurations is less than or equal to 255 commands, only the configurations that are not available are applied to the node that is rejoining the cluster. The operational state of the node remains unaffected.