Este contenido no está disponible en el idioma seleccionado.
5.6. Configure the MessagingClusterHealthMBean
The
MessagingPostOffice
uses the KeepOldFailoverMode
and NodeStateRefreshInterval
parameters to control the interval during which a node must update timestamps to remain part of a cluster. These parameters do not control how a node responds to losing connection to the database.
If a node loses connection to the database, it can not update the timestamp information available to the
MessagePostOffice
. Even though the node may still be healthy, the cluster has no way of determining the actual state of the node. As a result, the cluster assumes the node has failed and duplicate message delivery can occur.
The
MessagingClusterHealthMBean
MBean is responsible for monitoring the node state, and stopping and starting a node when it loses connection to the database. When a node is shunned from the cluster and subsequently loses database connectivity, the MBean shuts down the node immediately. The MBean monitors the JGroups status and database status while the node is unavailable, and will restart the node when it detects JGroups and the database connection is restored to normal operation.
To enable this functionality, declare the
MessagingClusterHealthMBean
MBean in the ServerPeer
MBean as an optional-attribute-name
attribute of the <depends>
directive.
After declaring the
MessagingClusterHealthMBean
MBean in the ServerPeer
MBean, you need to add the MessagingClusterHealthMBean
MBean's configuration.