Product Documentation

Configuring redundancy for nodegroups

Note

Supported from NetScaler 10.5 Build 52.1115.e onwards.

Nodegroups can be configured such that when one nodegroup goes down, another nodegroup can take over and process traffic. For example, when a nodegroup NG1 goes down, NG2 takes over.

Note

This functionality can be used to configure datacenter redundancy where each nodegroup is configured as a datacenter.

To achieve this use case, cluster nodes must be logically grouped into nodegroups, where some nodegroups must be configured as ACTIVE and others as SPARE. The active nodegroup with the highest priority (that is, the lowest priority number) is made operationally active and therefore serves traffic. When a node from this operationally active nodegroup goes down, the node count of this nodegroup is compared with the node count of the other active nodegroups in order of their priority. If a nodegroup has a higher or equal node count, that nodegroup is made operationally active. Else, the spare nodegroups are checked.

Note

  • Only one state-specific nodegroup can be active at a given point in time.
  • A cluster node inherits the state of the nodegroup. So, if a node with “SPARE” state is added to nodegroup with state as “ACTIVE”, the node automatically behaves as an active node.
  • The preemption parameter that is defined for the cluster instance decides whether the initial active nodegroup will take control when the it comes up again.
  • A spare node group can take up a node group and host active traffic when an active node group goes down.

The following figure shows a nodegroup setup that has nodegroup redundancy defined. NG1 is initially the active nodegroup. When it loses one of the nodes, the spare nodegroup (NG3) with the highest priority starts serving traffic.

Figure 1. Citrix ADC cluster with nodegroup redundancy configured

localized image

Configuring redundancy for nodegroups

  1. Log on to the cluster IP address.

  2. Create the active nodegroup and bind the required cluster nodes.

    > add cluster nodegroup NG1 -state ACTIVE
    > bind cluster nodegroup NG1 -node n1
    > bind cluster nodegroup NG1 -node n2
    > bind cluster nodegroup NG1 -node n3
    
  3. Create the spare nodegroup and bind the requisite nodes.

    > add cluster nodegroup NG2 -state SPARE -priority 20 
    > bind cluster nodegroup NG2 -node n4
    > bind cluster nodegroup NG2 -node n5
    > bind cluster nodegroup NG2 -node n6
    
  4. Create another spare nodegroup and bind the requisite nodes.

    > add cluster nodegroup NG3 -state SPARE -priority 10
    > bind cluster nodegroup NG3 -node n7
    > bind cluster nodegroup NG3 -node n8
    > bind cluster nodegroup NG3 -node n9
    

Configuring redundancy for nodegroups