|
|
|
|
|
The cluster connector is an optional agent that handles communication between a VCS cluster and the management server. If used, the cluster connector runs on the cluster and always initiates the connection from the cluster to the management server. Using the cluster connector is the alternative to using a direct connection.
You must use the cluster connector under the following conditions:
On the cluster, the cluster connector communicates with the VCS engine, HAD. The cluster connector runs on one system, or node, of the cluster, but is installed on each node and is configured for failover.
The cluster connector buffers VCS information. One function of this buffer is to provide some protection against loss of data if the cluster connector loses the connection to the management server. When the connection reestablishes, the management server retrieves the data in the cluster connector buffer. This data is the cluster status and event information that was generated during the disruption. The management server subsequently makes that information available in the Cluster Management Console.
The buffer also enables the cluster to become available for management much faster after a disruption. When the cluster reconnects, it needs to transmit only those events that transpired while the connection was down, which are stored in the buffer.
Note
The cluster connector buffer has a size limit. The ClusterConnectorConfig agent configures this limit using the value of the MessageBufferSize attribute. The MessageBufferSize attribute is part of the resource type definition for the agent. If status and event information generated during the disruption exceeds this limit, a partial loss of data occurs.
See ClusterConnectorConfig agent.
Although the cluster connector is designed to initiate the connection to the management server, it does not require the existence of intervening firewalls. You may choose to use it for the buffering feature alone or out of personal preference.