16.3.4 Cluster settings
Specify the following settings for a cluster:
-
Creating a script file
-
Cluster software environment settings
- Organization of this subsection
(1) Creating a script file
Create script files for starting and stopping Management Server, Administration Agent, and Application Server. You must also determine the cluster monitoring method in the script file. For details on creating the script files, see the OS manual in use.
This subsection describes the monitoring targets and monitoring methods of node switching for the host management model.
(a) Monitoring targets
-
Management Server
-
Administration Agent
-
Logical server #
- #
-
The node is switched when the logical server status becomes "abnormal termination" (the automatic restart frequency is exceeded or when 0 is specified as the automatic restart frequency and a failure is detected). Specify the following key in mserver.properties (Management Server environment setup file):
com.cosminexus.mngsvr.logical_server_abnormal_stop.exit = true
If automatic restart is specified for Management Server, node switching is not executed during this time. For executing the node switching during this time, do not specify automatic restart for Management Server.
In a configuration in which two or more executing nodes exist, do not specify true for the com.cosminexus.mngsvr.logical_server_abnormal_stop.exit key. If you do so, multiple systems of these executing nodes might start on the standby node.
(b) Monitoring methods
This subsection describes the monitoring methods to be implemented in node switching for the host management model.
Use one of the following methods according to the operation or environment used:
-
Process monitoring
This method does not place too much load on the monitoring process.
-
Process monitoring and health check
The monitoring accuracy is higher than the method of monitoring only the processes, but this method places some load on the monitoring process.
To monitor a logical server, see the notes described in (a) Monitoring targets.
-
Process monitoring
This method monitors the existence of Management Server processes. The following table describes the names of the monitored processes:
Table 16‒3: Monitored processes Monitoring target
Process name
In Windows
In UNIX
Management Server
mngsvr.exe
cjstartsv#1, #2
Administration Agent
adminagent.exe
adminagent
Logical server
mngsvr.exe
cjstartsv#1
- #1
-
The cjstartsv process has the same name as the process name of the J2EE server. You identify the process of Management Server by the server name.
- #2
-
The command line name of Management Server begins with the following character string:
/opt/Cosminexus/CC/server/bin/cjstartsv cosmi_m
cosmi_m is the default server name. If Management Server is set up with a server name other than cosmi_m, change the server name.
-
Health check
This method monitors the existence of Management Server processes and whether the existing processes are running. Confirm by executing the following commands in Management Server:
-
When monitoring Management Server and logical server
mngsvrutil command
-
When monitoring the Administration Agent
adminagentcheck command
-
(2) Cluster software environment settings
For details on the cluster software environment settings, see the cluster software documentation in use.