Hitachi

JP1 Version 12 JP1/Automatic Job Management System 3 System Design (Configuration) Guide


4.3.2 Activating multiple scheduler services concurrently

A scheduler service is a control unit for managing root job groups, and is configured within the JP1/AJS3 services on a manager host. One or more root job groups must be defined at the highest level of the scheduler service hierarchy. The name of a root job group is a forward slash (/).

The following figure shows a configuration with multiple scheduler services activated.

Figure 4‒2: Using multiple scheduler services

[Figure]

When you use multiple scheduler services, the root job groups can be managed in different scheduler services. However, regardless of how many scheduler services are used, resources might be stretched in a system where more than 4,000 jobnets and jobs need to be managed (defined, registered for execution, and monitored) in one scheduler service. If your system has several CPUs, consider running multiple scheduler services concurrently.

For example, you could activate multiple scheduler services and divide work tasks among them. This would enable CPU resources to be utilized efficiently by each scheduler service, and each scheduler service would be able to execute its own work tasks (jobnets and jobs) independently of the others. Test runs could also be performed without affecting other operations.

You must also consider running multiple scheduler services when a very large number of jobnets need to be registered for execution. Taking account of processing performance, divide the jobnets into about 4,000 per scheduler service. Splitting up jobnets per work task can also help to reduce jobnet size.

A maximum of 20 scheduler services can be activated concurrently on one computer, including both the physical host and the logical hosts. However, depending on the system environment, if you specify more than a certain number of concurrent scheduler services in the environment settings, some services might not start.

To activate multiple scheduler services concurrently, define each one separately in the environment settings for each logical host. For the required settings, see 6.1.1 Settings for starting multiple scheduler services in the JP1/Automatic Job Management System 3 Configuration Guide (for Windows systems) or 15.1.1 Settings for starting multiple scheduler services in the JP1/Automatic Job Management System 3 Configuration Guide (for UNIX systems).

To start or stop scheduler services individually, execute the jajs_spmd or jajs_spmd_stop command with the -n jajs_schd option specified. You can execute these commands while JP1/AJS3 services are active. For details, see 7.5 Starting and stopping only the scheduler service in the JP1/Automatic Job Management System 3 Administration Guide.

Organization of this subsection

(1) Activating multiple scheduler services on a Windows host

Depending on the system environment, an error might occur when more than a certain number of schedulers are activated concurrently. In this case, an error message reports that the application was not initialized correctly.

This error occurs when there are insufficient system resources (desktop heap shortage).

In JP1/AJS3, each scheduler service activates many control processes. When multiple scheduler services are activated concurrently, usage of the desktop heap increases proportionately and shortages might occur. In such cases, take the following measures.

(a) Use the desktop heap area specific to JP1/AJS3

To use the desktop heap area specific to JP1/AJS3, change the JP1/AJS3 services account to a user account. The system account is set by default for JP1/AJS3 services.

For details about this setting, see 4.2.3(1) Changing the account for services provided by JP1/AJS3.

(b) Adjust the desktop heap size by editing the registry

Edit the Windows registry to adjust the desktop heap size. For details about the procedure, see the relevant articles on Microsoft's Help and Support website. Usage of the desktop heap depends on your system environment. Be sure to consider the consequences before changing the heap size.

(2) Activating multiple scheduler services on a UNIX host

Increasing the number of scheduler services will increase system resource usage in direct proportion. If sufficient system resources cannot be allocated for the number of scheduler services you are using, an error might occur. Set the number of scheduler services to be activated concurrently once you have estimated the memory requirements, disk space requirements, kernel parameters, and other system resources needed, as described in 3. Estimates.