Job Management Partner 1/Automatic Job Management System 3 System Design (Configuration) Guide
This subsection describes the databases used by the scheduler control, job execution control, and agent management control.
For details about the database for QUEUE jobs and submit jobs used by the submit job control, see 7.1.3 Database for QUEUE jobs and submit jobs.
- Organization of this subsection
- (1) Scheduler database
- (2) Agent management database
The following table describes the information stored in the scheduler database used by the scheduler control and job execution control.
Table 2-22 Tables in a scheduler database
Table name#1 Contents Row length#2
(bytes)Variable-length rows#2
(bytes)Number of indexes Index length#2
(bytes)Definition
tableAJSUNIT Unit management 408 None 2 Primary keys: 4 Sub-keys: 36 AJSARROW Unit relationship definition 52 None 1 12 AJSBODY Unit definition 83,268 4 to 64,000 1 4 AJSSCH Schedule definition 148 4 to 8,192 1 8 AJSCAL Calendar definition 1,540 None 1 8 AJSRELS Release definition 1,180 None 3 Primary keys: 36 Sub-keys: 8 Sub-keys: 8 Execution
tableAJSSTAT Unit status 1,540 None 1 12 AJSENTRY Registration for execution 764 None 1 8 AJSGEN Generation management 124 4 to 96,000 1 8 AJSPERF Statistics 120 None 2 Primary keys: 4 Sub-keys: 8 AJSID ID management 76 None 1 8
- #1
- Default names in the JP1/AJS3 environment settings.
- For an embedded database, the index has a table name followed by INDEX1 for the primary key (including tables with no sub-keys), and INDEX2 and INDEX3 for the sub-keys.
- #2
- The row length given for a variable-length table is the size of the fixed-length portion at the beginning of each record.
- Note that in addition to this size, the table region on the disk includes a management area and therefore does not match the table size given above.
- A JP1/AJS3 database has a maximum variable-row length of 64,000 bytes, and contains one field for data plus another variable-length field. The variable-length field (maximum 64,000 bytes) has two VARCHAR fields. If the variable-length data to be stored in one of the two fields does not exceed 32,000 bytes, zero-length data (SQLLEN = 1) is set in the second VARCHAR field.
The following describes the number of records in each table and how the variable-length records are used.
The tables in the scheduler database can be divided into tables for storing unit definitions (definition tables) and tables for storing job and jobnet schedules and results (execution tables).
The following describes the record count and variable-length records for each type of definition table.
- Unit management table (AJSUNIT)
- The number of records is the number of JP1/AJS3 units (total number of job groups, jobnets, and jobs).
- Unit relationship definition table (AJSARROW)
- The number of records is the number of relationships established among jobnets.
- Unit definition table (AJSBODY)
- The number of records is the total number of job definitions and jobnet definitions.
- Except for schedule information and calendar information, unit definition information including command statements and script file names is stored.
- Schedule definition table (AJSSCH)
- The number of records is the total number of jobnets plus the total number of scheduling rules defined for them. The table stores the common schedule information and scheduling rule information.
- The common schedule information (names of units referencing the calendar) and scheduling rule information (scheduling rules) are stored as variable-length data in the same format as the parameters in the unit definition file.
- Calendar definition table (AJSCAL)
- The number of records is the number of job groups plus the number of years defined in the calendar.
- Release definition table (AJSRELS)
- The number of records is the number of jobnets registered for release plus the number of times each jobnet has been registered for release. As JP1/AJS3 automatically deletes these records depending on the status of the jobnet registered for release, the number of records might be different from the calculated figure.
The following describes the record count and variable-length records for each execution table.
- Unit status table (AJSSTAT)
- The number of records necessary for each registered root jobnet is root-jobnet-and-total-number-of-nested-jobnets-and-jobs x (number-of-logs-to-keep + number-of-scheduled-generations-registered-for-fixed-execution + 1), where the number of logs to keep includes sub-generations created when start conditions are satisfied.
- Calculations are made for all registered root jobnets, and the sum becomes the total number of records.
- Registration for execution table (AJSENTRY)
- The number of records necessary for each registered root jobnet is number-of-logs-to-keep-for-the-root-jobnet + number-of-scheduled-generations-registered-for-fixed-execution, where the number of logs to keep includes sub-generations created when start conditions are satisfied. Calculations are made for all registered root jobnets, and the sum + 1 becomes the total number of records.
- Generation management table (AJSGEN)
- The number of records and amount of information stored as variable-length data differs depending on whether start conditions are used for the registered root jobnets.
- For a root jobnet without a start condition, an item of information with a length of ((number-of-logs-to-keep + number-of-scheduled-generations-registered-for-fixed-execution) + 1) x 8 bytes is stored as variable-length data.
- In the same way, for a root jobnet that has a start condition, an item of data with a length of ((number-of-logs-to-keep + number-of-scheduled-generations-registered-for-fixed-execution) + 1) x 8 bytes is stored as variable-length data. In addition, information about the number of sub-generations, calculated as number-of-sub-generations-created-every-time-a-start-condition-is-satisfied x 8 bytes is stored as variable-length data.
- Calculations are made for all registered root jobnets, and the sum becomes the total number of records.
- Statistics table (AJSPERF)
- The number of records is the total number of units registered for execution.
- ID management table (AJSID)
- The number of records is the number of records in all the active units.
(c) Supplementary notes about the scheduler database
If the scheduler service does not stop as planned due to a power failure or other reason, processing resumes when the service next starts, and the KAVS0211-W message is output to the Windows event log or syslog. Check the job status as jobs might have stopped before completion.
The agent management database is used by the agent management control. This database manages definition information about execution agents and execution agent groups. The following table describes the database tables in an agent management database.
Table 2-23 Tables in an agent management database
Table name Contents Row length#
(bytes)Variable-length rows
(bytes)Number of keys Number of indexes Index length (bytes) EXECAGT Execution agent definition 952 None 1 1 4 AGTEXECCNT Maximum concurrent jobs definition 20 None 1 1 8 EXECAGTGRP Execution agent group definition 696 None 1 1 4 LINKAGT Connected execution agents definition 20 None 1 1 8 The number of records in each table is as follows:
- Execution agent definition (EXECAGT)
- The number of records is the number of defined execution agents.
- Maximum concurrent jobs definition (AGTEXECCNT)
- The number of records is number-of-defined-target-agents x 16.
- Execution agent group definition (EXECAGTGRP)
- The number of records is the number of defined execution agent groups.
- Connected execution agents definition (LINKAGT)
- The number of records is the total number of execution agents connected to each execution agent group.
Copyright (C) 2009, 2010, Hitachi, Ltd.
Copyright (C) 2009, 2010, Hitachi Solutions, Ltd.