This subsection describes how to set up Performance Management to run in a cluster system.
Setup must be performed first on the executing node and then on the standby node.
indicates a task to be performed on the executing node.
indicates a task to be performed on the standby node.
indicates a setup item that is required depending on the environment or an optional setup item used when the default is to be changed.
To perform integrated management of PFM - RM for Oracle using PFM - Manager and PFM - Web Console, you must register PFM - RM for Oracle with PFM - Manager and PFM - Web Console.
You will need to register PFM - RM for Oracle in the following cases:
Registration is performed separately for PFM - Manager and PFM - Web Console. The registration procedure is the same as the registration procedure in a non-cluster system.
For details about the procedure, see 2.2.4(2) Register PFM - RM for Oracle.
Make sure that the shared disk is mounted. If the shared disk is not mounted, use the cluster software or the volume manager to mount it.
Execute the jpcconf ha setup command to set up the logical host environment. When you execute the command, the necessary files are copied to the shared disk, the logical host definition is set up, and the logical host environment is created.
To set up the logical host environment:
jpcconf ha setup -key RMOracle -lhost jp1-halora -d /jp1
jpcconf ha list -key all
Execute the jpcmgrhost define command to specify which PFM - Manager is to manage PFM - RM for Oracle.
jpcconf mgrhost define -host jp1-hal -lhost jp1-halora
Create an Oracle account that has specific system privileges to collect performance data for the Oracle Database monitored by PFM - RM for Oracle.
For details about how to create an Oracle account, see 2.2.4(3) Create an Oracle account to be used in PFM - RM for Oracle.
This setup task is unnecessary when you use the sys account.
You execute the jpcconf inst setup command to set up an instance environment for PFM - RM for Oracle.
The setup procedure is the same as the setup procedure in a non-cluster system, except that, in a cluster system, you must specify the logical host name in the -lhost option when executing the jpcconf inst setup command.
In a cluster system, the jpcconf inst setup command is executed in the following format:
jpcconf inst setup -key RMOracle -lhost logical-host-name -inst instance-name
Although an example of interactive command execution is shown here, the jpcconf inst setup command can be also executed non-interactively. For details about the jpcconf inst setup command, see the chapter that describes commands in the manual Job Management Partner 1/Performance Management Reference.
As the agent log output folder (the value of log_path), specify the path of a folder that is on the shared disk.
For details about other settings and procedures, see 2.2.4(4) Set up an instance environment.
Execute jpcconf target setup command to specify which Oracle host for PFM - RM for Oracle to monitor.
The setup procedure is the same as the setup procedure in a non-cluster system, except that, in a cluster system, you must specify the logical host name in the -lhost option when executing the jpcconf tatget setup command.
In a cluster system, the jpcconf target setup command is executed in the following format:
jpcconf target setup -key RMOracle -inst instance-name -target monitoring-target-name -lhost logical-host-name
Although an example of interactive command execution is shown here, the jpcconf target setup command can be also executed non-interactively. For details about the jpcconf target setup command, see the chapter that describes commands in the manual Job Management Partner 1/Performance Management Reference.
For details about other settings and procedures, see 2.2.4(4)(b) Set the monitoring target.
At this point, set up any other Performance Management programs, such as PFM - Manager or PFM - RM, on the same logical host.
For details about how to set up these products, see the chapters on setup and operation in a cluster system in the Job Management Partner 1/Performance Management User's Guide, and the chapter on operation in a cluster system in the manuals for the applicable version of PFM - RM.
This setup task is necessary only when you change the network settings to match the configuration in which Performance Management is used.
The following are the two network setting items:
Performance Management outputs its operating status to a set of proprietary log files called a common message log. By default, the common message log consists of two 2,048 KB files. Perform this setting if you want to change the default file size.
For details, see the chapter on installation and setup in the Job Management Partner 1/Performance Management Planning and Configuration Guide.
Perform this setting only if you want to change the folders where the database of performance data managed by PFM - RM for Oracle is saved, backed up, exported, or imported.
For details, see 2.6.1 Changing the storage location of performance data.
This setting is required to output action log when alarm occur. The action log stores history information that is output in conjunction with the alarms for thresholds related to system load and other conditions.
For details about how to set up the action log, see K. Outputting Action Log Information.
Once you have created the logical host environment for PFM - RM for Oracle, export the environment definition as a file. In the export process, a file containing the collective definition information for all Performance Management programs set up on that logical host is output. Export the environment definition only after you have set up all additional Performance Management programs on the logical host.
To export the environment definition of the logical host:
jpcconf ha export -f lhostexp.txt
Copy the file you exported in (13) Export the logical-host environment definition file from the executing node to the standby node.
Complete setup by unmounting the file system. Note that if you intend to continue using the shared disk, you do not need to unmount the file system at this point.
On the standby node, import the exported file you copied from the executing node.
Use the jpcconf ha import command to set up the environment definition for the Performance Management programs of the logical host that you created on the executing node so that they to run on the standby node. If more than one Performance Management program was installed on the logical host, the definition information for all of the programs is imported in batch form.
Note that the shared disk does not need to be online when you execute this command.
To import the logical-host environment definition file:
jpcconf ha import -f lhostexp.txt
jpcconf ha list -key all
If you intend to use Performance Management programs in a logical host environment, make sure that the programs are registered in the cluster software. Also, set up the environment so that the Performance Management programs are started and stopped based on instructions from the cluster software.
This subsection describes the settings to be specified to register PFM - RM for Oracle in the cluster software.
In UNIX, you will typically need to register control methods for Start, Stop, Process monitoring, and Forced stop in the cluster software.
The following table shows how to set these control methods for PFM - RM for Oracle.
Table 3-5 Control methods for PFM - RM for Oracle registered in the cluster software
Item | Description |
---|---|
Start | Start PFM - RM for Oracle by executing the following commands: /opt/jp1pc/tools/jpcspm start -key AH -lhost logical-host-name |
Stop | Stop PFM - RM for Oracle by executing the following commands: /opt/jp1pc/tools/jpcspm stop -key RMOracle -lhost logical-host-name -inst instance-name If the service has stopped due to a failure, the jpcspm stop command returns 3. In this case, assume that the command terminated normally, because the service has stopped. For cluster software that evaluates the execution result from a return value, convert the return value to 0, for example. |
Process monitoring | Use the ps command to check whether the monitored processes are running: ps -ef | grep "process-name logical-host-name" | grep -v "grep monitored-process" Monitored processes are as follows:jpcagt1, agt1/jpcsto, jpcah For details about process names, see the chapter that describes configuration and operation in a cluster system in the Job Management Partner 1/Performance Management User's Guide. Hitachi also recommends that you provide a control method that allows the cluster software to temporarily stop monitoring. This can be used when temporarily stopping Performance Management, for example, during system maintenance (thereby stopping the cluster software from monitoring files that are undergoing maintenance). |
Forced stop | To perform a forced stop, execute the following command: /opt/jp1pc/tools/jpcspm stop -key all -lhost logical-host-name -kill immediate As the service key of the first argument, you can specify only all.
|
Check whether the cluster software is operating correctly by using it to issue start and stop requests to Performance Management programs on each node.
After setting up the Performance Management programs, use PFM - Web Console to set up the environment for the programs. You will then be able to display reports on the operating status of monitoring targets, and notify users whenever a problem occurs.
For details about setting up the environment for Performance Management programs, see the chapters on setup and operation in a cluster system in the Job Management Partner 1/Performance Management User's Guide.