Job Management Partner 1/Automatic Job Management System 3 System Design (Configuration) Guide

9.3 Notes on the JP1/AJS3 database

Note the following points in regard to the JP1/AJS3 database:

High-priority processes (the JP1/AJS3 scheduler services) can monopolize the CPU, causing JP1/AJS3 processing performance to decline. However, setting a lower priority for the JP1/AJS3 scheduler services, and a higher priority for the ajsmonsvr process and commands that operate on units, can also cause start delays in jobs and jobnets. When executing commands for operating on units, make sure you use them in an environment where they have the same priority as the scheduler services.
This can be more of an issue in UNIX, depending on the set of nice values. Note the following in regard to the nice value at process startup:

When executing a command that operates on units
Execute the command with the same nice value as the JP1/AJS3 service.

When executing, as a JP1/AJS3 job, a command that operates on a unit (including a command that operates on a unit and is executed within a script specified as a job)
Change the jobnet's Priority parameter from the default (None) to 3. If you start the JP1/AJS3 service with a nice value of 20, +20 applies to the default setting of None and corresponds to a nice value of 39 (the capped maximum), resulting in the lowest job priority. With a setting of 3, however, +0 applies and corresponds to a nice value of 20. This means that the job will be executed with the same nice value as the JP1/AJS3 service, and is less likely to monopolize the CPU.

When the embedded database is used for the scheduler services, the system can be operated for a very long time, as long as you periodically execute the ajsembdbreclaim command. If the system is changed significantly (if you delete or cancel registration of a large-scale jobnet, for example), check the number of unused segments by executing the ajsembdbstatus command. The segment reuse rate might be down if the percentage of used segments exceeds 80% after prolonged system operation. (Under the embedded database's standard setup by the JP1/AJS3 setup script, message KFPH00211-I is output to the Windows event log or UNIX syslog when usage reaches 80%, 90%, and 100%.) If the percentage of used segments exceeds 80%, you will need to reorganize the database.

All space characters at the end of a comment describing a unit are ignored. Do not enter any spaces after a comment.

[Contents][Back][Next]

[Trademarks]