Cluster
persistence,
caching, and
multicast are configured in the
cluster
node configuration file.
When
you configure a cluster,
enable
persistence
to save projects
and workspaces when the cluster shuts down.
All nodes in a cluster must point to the same persistence
directory, which
defaults to
ESP_HOME/storage.
Note: If the nodes in a cluster are on different machines, the persistence folder must
be on a shared disk.
When
persistence is disabled, you lose all your projects when the last manager node in the
cluster shuts down; therefore, in a production system, Sybase recommends that you enable
persistence.
Caching and Multicast
The cluster cache is an in-memory distributed cache used for internal sharing of
cluster state and configuration. Manager nodes are members of the cache, while
controller nodes are clients of the cache. Cache properties must be configured for
all nodes, but you do not need to provide port information for controller-only
nodes. For all
nodes added to this cluster, configure the same name and password in the Cache
section of <node-name>.xml.
Managers discover and join an existing cluster cache in one of two ways:
- Multicast – each manager node broadcasts its connection details so other managers
can join if they have the correct credentials. Recommended for testing
environments where all nodes in the cluster are on the same subnet.
Multicast does not work well when cluster nodes are on different subnets
or when multiple clusters use the same names and ports (usually the
defaults).
- Direct connect – each manager node uses the host name and port information in the
Managers section of its <node-name>.xml file to
discover and join the cache. Recommended for production
environments.
If you enable multicast, enable it on all nodes. A cluster cannot function properly
unless all its nodes use the same form of communication for caching.