18.7 Starting and stopping the mutual node switching system (In UNIX)
This section describes how to start and stop the system when you use the HA monitor-based mutual node switching system. This section also describes the starting and stopping procedure when maintenance is performed for the J2EE server and batch server after the start of operations. Note that the HA monitor can only be used in AIX or in Linux.
To use the mutual node switching system, you must first specify the necessary environment settings such as preparing the two types of hosts as the active node and the spare node, and registering the script for monitoring, starting, and stopping the Administration Server to be monitored by the HA monitor. For details on the setup methods, see the description related to the HA monitor settings in 18.5.4 HA monitor environment settings.
The following figure shows a configuration example of the mutual node switching system. This subsection describes how to start and stop the mutual node switching system based on this system configuration.
|
From the servers file definitions (server environment settings in the HA monitor) in each Application Server in the figure, the following table lists the definition items related to the starting and stopping of the system.
Operand |
Description |
Specified value |
|
---|---|---|---|
Active node 1 (Virtual host 1) |
Spare node 2 (Virtual host 2) |
||
alias |
Server identification name |
AP1 |
AP2 |
initial |
Server starting method in the HA monitor |
online |
standby |
Operand |
Description |
Specified value |
|
---|---|---|---|
Spare node 1 (Virtual host 1) |
Active node 2 (Virtual host 2) |
||
alias |
Server identification name |
AP1 |
AP2 |
initial |
Server starting method in the HA monitor |
standby |
online |
For details on the servers file definitions, see 18.5.6 Server-compliant environment settings.
The following commands are provided by the HA monitor. For details, see the manual Reliable System Monitoring Functionality HA Monitor.
-
monbegin (Starting a server that does not have an interface with the HA monitor)
-
monend (Communication to stop the executing server that does not have an interface with the HA monitor)
-
monswap (Planned node switching)
- Important note
-
If node switching occurs due to a failure in the Management Server or Administration Agent, the logical server processes might remain in the failed host. If processes remain in the host, the logical server cannot be started when node failback occurs, therefore, all the processes must be stopped before node failback occurs. In addition, implement operations for starting the business correctly in advance.
- Organization of this section