Cluster Manager
The Cluster Manager provides a set of properties for communication between the elements in the cluster. It integrates new elements and notifies the modules installed on each element when an element is removed. This manager runs on both servers and dispatchers.
Dispatcher
Property |
Description |
Default Value |
barrier.dependent |
Determines if the startup and shutdown barriers are dependent. If the property value is "true", there can be only one starting or stopping element in the box/cluster at the moment. If it is "false", there can be only one starting and one stopping element. |
false |
barrier.shutdown |
Determines the serialization kind of the element's shutdown (similar to barrier.startup). |
none |
barrier.startup |
Determines the serialization kind of the element's startup. The following values are available: "none" – no serialization; "box" – serialization in the boundaries of the box, that is, the nodes in one box will be started one by one; "cluster" – serialization in the boundaries of the whole cluster. |
none |
barrier.timeout |
An integer that specifies a timeout in milliseconds. Within this timeout, the element is forced to recheck if it can pass the barrier.
The system uses this property for optimization purposes only. We do not recommend that you change it.
|
10000 |
element.clusterId |
The identification number of the cluster element. The ID is unique for the cluster.
Do not modify this property. |
Assigned at installation time |
element.groupId |
The group ID of the element.
Do not modify this property. |
Assigned at installation time |
element.name |
Specifies the name of the cluster element. |
Dispatcher <n> |
element.type |
Specifies the type of the cluster element (server or dispatcher).
Do not modify this property. |
dispatcher |
ms.backup.message.number |
The length of the internal message backup queue.
Do not modify this property. |
10 |
ms.confirmation.timeout |
The timeout period (in milliseconds) that a joining cluster node waits for the confirmation replies from the cluster participants. |
30000 |
ms.eyecatcher.control |
Specifies if the eyecatcher control is activated.
Do not modify this property. |
false |
ms.host |
This is the host (IP address), where the Message Server is running. |
|
ms.keepalive |
If there has been no communication between a cluster node and the message server for the specified timeout period, the node triggers a liveliness check (ping/pong protocol). The value is in milliseconds. |
20000 |
ms.message.pool.size |
Defines the size of the message library pool. The size of this pool can be changed for optimization purposes. |
10000 |
ms.notification.timeout |
Defines the timeout (in milliseconds) for receiving an acknowledgement by the message server that the message has been delivered.
Do not modify this property. |
180000 |
ms.port |
The port on which the Message Server listens for connections. |
3601 |
ms.reconnect.timeout |
Specifies the timeout (in milliseconds) after which the node is no longer able to reconnect to the message server. |
180000 |
session.buffer.size |
The size of the I/O stream buffers that wrap the session socket I/O stream. |
2048 |
session.message.queue.size |
Defines the size of the session layer queue. The size of this queue can be changed for optimization purposes. |
256 |
socket.SO_KEEPALIVE |
Enables and disables the socket “keep alive” option. This property has a Boolean value. If it is false, the socket “keep alive” option is disabled. |
false |
socket.SO_LINGER |
Enables and disables a socket linger with the specified time in seconds. The maximum timeout value is platform specific. “-1” implies that the option is disabled. The setting only affects the socket close. |
-1 |
socket.SO_RCVBUF |
Sets the received buffer size for this socket.
This property has the same use as the socket.SO_SNDBUF property described above. |
65536 |
socket.SO_SNDBUF |
Sets the send buffer size for this socket. This option is used by the platform's networking code as a hint for the size to set the underlying network I/O buffers. Increasing the buffer size can increase the performance of network I/O for high-volume connection, while decreasing it can help reduce the backlog of incoming data. |
65536 |
socket.SO_TIMEOUT |
Enables and disables the socket timeout with the specified value in milliseconds. The timeout must be > 0. A timeout of zero is interpreted as an infinite timeout, that is, the timeout is disabled. |
0 |
socket.TCP_NODELAY |
This property takes a Boolean value that specifies whether Nagle’s algorithm is disabled or enabled for this socket. Setting socket.TCP_NODELAY to “true” improves system performance, especially if multiple small portions of data are sent through the socket. |
true |
Server
Property |
Description |
Default Value |
barrier.dependent |
See above. |
false |
barrier.shutdown |
See above. |
none |
barrier.startup |
See above. |
none |
barrier.timeout |
See above. |
10000 |
element.clusterId |
See above. |
Assigned at installation time |
element.groupId |
See above. |
Assigned at installation time |
element.joinPort |
The server listens for connections on this port. |
Assigned at installation time |
element.name |
See above. |
Server <n> |
element.type |
See above. |
server |
lazy.autoclose.timeout |
Specifies the time through which the lazy connection will be open after the last large group of message is transported. After this timeout the Cluster Manager closes the connection. The time is in milliseconds. The maximum for closing a socket is between 5 and 10 seconds. |
5000 |
lazy.exclusive.list |
A list of services for which if the Cluster Manager has to transfer messages from or to them, it always transfers them using the lazy communication mechanism without considering the lazy.threshold value. |
jmx, jms_provider |
lazy.message.pool.size |
Specifies the size of the pool where the currently unused message objects are stored. This message objects are used by the messaging system for internal use only. The value must be equal or greater than “0”. |
10000 |
lazy.threshold |
Specifies the size of the messages that must be sent during a time interval equal to the value set in the lazy.time.piece property so that if this size is exceeded a lazy communication is established. The value must be equal or greater than “0”. If the value is “0”, all messages will be transferred through a lazy connection. |
100000 |
lazy.time.piece |
Specifies the time interval in milliseconds for which the quantity of messages defined by the lazy.threshold property have to be exchanged between two parties in order to open a lazy connection. If two parties open a connection and succeed in exchanging more messages than the size defined by the lazy.threshold property for the time defined in this property, then a lazy connection will be opened. |
1000 |
lazy.transparent.switch |
If set to true, specifies that the value of the lazy.threshold property is taken into account when establishing a lazy communication. If set to false, only messages from or to the services specified in the lazy.exclusive.list property are sent through the lazy communication mechanism. |
false |
ms.backup.message.number |
See above. |
10 |
ms.confirmation.timeout |
See above. |
30000 |
ms.eyecatcher.control |
See above. |
false |
ms.host |
See above. |
Depends on the settings made at installation |
ms.keepalive |
See above. |
20000 |
ms.message.pool.size |
See above. |
10000 |
ms.notification.timeout |
See above. |
180000 |
ms.port |
See above. |
3604 |
ms.reconnect.timeout |
See above. |
180000 |
session.buffer.size |
See above. |
2048 |
session.message.queue.size |
See above. |
256 |
socket.SO_KEEPALIVE |
See above. |
false |
socket.SO_LINGER |
See above. |
-1 |
socket.SO_RCVBUF |
See above. |
65536 |
socket.SO_SNDBUF |
See above. |
65536 |
socket.SO_TIMEOUT |
See above. |
0 |
socket.TCP_NODELAY |
See above. |
true |
See also:
Cluster Manager Configuration in the Administration Manual
Communication Management in the Administration Manual