How do I configure the CRX cluster, and what configuration options do I have for CRX 1.4.1 or 1.4.2?
The first cluster node that is started will act as the master node. The master node will open TCP listener(s), and cluster nodes that are started later will automatically try to connect to this master node. One of the requirements of a CRX cluster is that the cluster nodes can connect to each other over TCP/IP.
Note: If another cluster node is configured as the preferred, it will automatically become the master as soon as it is started.
By default, a cluster nodes reads the journal and update its state (including its cache) every 5 seconds (5000 milliseconds).
This setting only applies if the cluster nodes is reading. If the cluster node is writing, the journal is read immediately.
The CRX file journal is configured as follows:
<Journal class="com.day.crx.core.journal.FileJournal"> <param name="<Parameter>" value="<Value>" /> ... </Journal>
|maximumSize||The default for CRX 1.4.x is 1048576 bytes. The value 104857600 is recommended for even for CRX 1.4.x|
When CRX / CQ starts, it automatically creates one file called
listener.properties for each workspace and one file for the journal. Those files are managed by the system and normally do not need to be changed or deleted. When CRX / CQ is not running, those files can be deleted.
This files contain the current cluster configuration information, including IP address and port of the current master in the cluster. These TCP/IP connections are used to transfer data between cluster nodes (sometimes called 'cluster interconnect'). They are also used for automatic master election.
Even if CRX / CQ is only running on one machine, the cluster interconnection is already used, but only on startup (using a local connection).
The CRX cluster configuration is similar to that of Apache Jackrabbit. For details, see http://wiki.apache.org/jackrabbit/Clustering
CRX 1.4.1 and 1.4.2