Cluster Management consists of the following:
Before you begin, select the server that will be your "master" server; it holds the authoritative data and is the only server capable of adding peer nodes ("slaves") to the cluster.
To create the cluster, access the master server, add the master/slave pair to it and apply a weighting factor for load balancing. Then access the slave server, add the master/slave pair to it, synchronize the system databases, and apply a weighting factor. If you have another MBG you wish to add to the cluster, follow a similar procedure to add it: First add the peer (slave) to the master and then add the master/peer (slave) pair to the peer server.
To create the cluster and add the master/slave pair to the master server:
To add the master/slave pair to the slave server and synchronize the system databases:
Wait for the master node to synchronize its database with the slave, which can take five minutes for newly created MBGs and up to 30 minutes for existing MBGs with large databases. When synchronization is complete, the State field on the master displays, "In sync with peer node" and the State field on the slave displays "In sync with master node."
Note: By default, the slave server has a cluster weight of zero (0). If you fail to update this value, the slave server will not participate in load balancing, even if all other nodes in the cluster fail.
To add a peer (slave) node to the master server for an existing cluster:
To add the master/peer (slave) pair to the peer server for an existing cluster:
Wait for the master node to synchronize its database with the peer (slave), which can take up to 30 minutes for existing MBGs with large databases. When synchronization is complete, the State field on the master displays, "In sync with peer node" and the State field on the peer (slave) displays "In sync with master node."
By default, the new node has a cluster weight of zero. If you fail to update this value, the new node will not participate in load balancing.
Failure of the master server is not service affecting. Any server in the cluster can raise an SNMP alarm to warn of outage or can take ownership of the cluster.
After a cluster is configured, the Clustering status frame provides the following information:
Cluster status |
|
Cluster Status |
The status of the local node: whether it belongs to a cluster, and whether it is the master node or a slave node. |
Manage Cluster |
Master:
|
Slave:
|
|
Zone status |
|
Zone name |
The name of this cluster zone. By default, the system has one zone, called "Default", but you can configure other cluster zones and then direct devices to use them. You can also rename the "Default" zone. After configuring a cluster zone, you should select a backup zone to act as a fallback in the event that the MBGs in the cluster zone become unavailable. |
Backup |
The name of the backup zone for this zone (if any). |
Node status |
|
Node ID |
The domain name/system name of nodes in the cluster. |
Node state |
The sync status of this node. |
Blade Version |
MBG software version installed on this node. |
Connected MiNet devices |
The number of phones currently configured in the cluster. |
Connected devices |
The number of phones currently connected to this node. |
Address |
The IP address this server uses to communicate cluster events with the cluster. |
Weight |
The weighting factor applied to this node. |
Connected |
üindicates that node is connected to the cluster. |
Data is automatically shared between master and slave servers in the cluster. Use the following procedures to manually synchronize the data.
Synchronizing a Peer with the Master
To manually re-synchronize a peer node with the master server in a cluster:
Synchronizing all Nodes in a Cluster
To manually re-synchronize all nodes in a cluster:
MBG clustering uses IP addresses to identify each node and to initiate cluster communications connections.
To change a node's IP address:
You can remove an individual node from a cluster or completely dissolve a cluster.
To remove a peer node from a cluster:
To completely dissolve a cluster:
To transform a slave into the master server of an existing cluster: