Clustering a Node Using Layer 3 Out-of-Band Manual Discovery

Starting in software version 5.1, a node residing on a different management subnet can join an out-of-band cluster using Layer 3 (L3) out-of-band manual discovery.

A node residing on a different IP subnet manually discovers the IP address of the current leader leader in clustering node relationship (formerly master) in the cluster and the IP address of the standby node. After discovering the IP addresses, the node residing on a different subnet establishes the connection with the current leader in the cluster. Once the node joins the cluster, it automatically receives a complete copy of the cluster’s database.

To allow a node on a different subnet to manually discover the primary leader IP address and the secondary standby IP address in the cluster, the auto-discovery of the cluster leader must be disabled. When the auto-discovery is disabled, the discovery process is manual. Also, the primary and secondary IP addresses must be configured for manual discovery.

Note:  The leader and the standby nodes must be reachable by the nodes residing on a different subnet.

For example:

(config) # no cluster leader auto-discovery

(config) # cluster leader address primary ip 192.168.1.52 port 60102

(config) # cluster leader address secondary ip 192.168.1.54 port 60102

Note:  When using manual discovery, the virtual IP address (vip) is not supported.

Manual discovery uses the primary IP address. When a leader fails and the standby is promoted to be the new leader, the node uses the secondary IP address and connects to the new leader. The node must discover the new leader within the specified timeout value.

For example:

(config) # cluster leader connect timeout 40

The default is 15 seconds. The values range from 10 to 120 seconds.

The nodes residing on a different subnet are not capable of becoming a leader or a standby node. They can only have the role of a normal node. The cluster leader preference assigned to these nodes are ignored.

Configuring Layer 3 Out-of-Band Manual Discovery

If the Mgmt IP address of all nodes reside on a different subnet and you want to put them into an out-of-band cluster, use the procedure in this section.

Refer to the following configuration example:

Step

Description

Command

1.    

Specify the cluster name, cluster ID, and cluster interface for all the nodes that will be part of the cluster.

Note:  Only eth0 interface is used for L3 out-of-band manual discovery.

(config) # cluster name Layer3-oob

(config) # cluster id Layer3-oob

(config) # cluster interface eth0

2.  

Verify that all the nodes have the same cluster information and that they are all running the same software version.

(config) # show cluster config

(config) # show version

3.  

Specify the cluster leader preference. For the node that is to be the leader, configure the highest preference, for example, 100. (For member node follower in clustering node relationship (formerly slave or non-master)s, the preference can be between 10 and 99.)

Note:  This command is not available for GigaVUE TA series.

(config) # cluster leader preference 100

4.  

Ping the Mgmt IP address of the node. If ping is good, enable the cluster on the leader first and then on the standby node.

(config) # ping mgmt-ip

(config) # cluster enable

5.

Verify that the cluster has formed.

(config) # show cluster global brief

(config) # show chassis

6.

On the leader, add the box ID of the other nodes in the cluster.

(config) # chassis box-id <box ID> serial-number <serial number>

7.  

For all other nodes residing on the different subnet, disable auto-discovery of the cluster leader. To enable L3 out-of-band manual discovery, configure the IP address used by the leader and the standby node.

(config) # no cluster leader auto-discovery

(config) # cluster leader address primary ip <IP address of leader>

(config) # cluster leader address secondary ip <IP address of standby>

8.  

Verify the configuration on a local node. Confirm the connectivity to the leader and the standby node.

(config) # show cluster config

(config) # ping mgmt-ip <IP address of leader>

(config) # ping mgmt-ip <IP address of standby>

9.  

Once the connectivity is confirmed, enable the cluster on the local node.

(config) # cluster enable

10.  

On the leader, repeat Step 5 and Step 6. Save the configuration.

(config) # write mem

11.  

Display cluster configuration

(config) # show cluster configured