Every job has a unique name.

Should a job with the same name be defined in one of the basic configurations, then this parameter can be used to change or supplement the settings made in that job.

order="yes" defines a job as being order controlled. The Job Scheduler will only start an order controlled job when an order for the job exists.

A script can check this attribute using the method.

A number of tasks can run in parallel from one job. This attribute specifies the maximum number of tasks for a job.

: only tasks="1", which only allows one task, makes sense in combination with an exclusive lock.

The Job Scheduler keeps this minimum number of tasks running. This allows order controlled tasks, which require a long time to initialize, to be held in waiting.

Note that the element must be large enough.

The Job Scheduler will start additional tasks when:

• it is starting
• a task has ended
• at the start of a new Period ()
• or the job is stopped by
• or all job tasks are continued with
• or the job is wakened by

• and fewer tasks than specified in min_tasks are running
• and the start is allowed according to the period
• and the job has either the pending or running states (that is, it is not being stopped or has not already stopped).

In order to prevent an overload, the Job Scheduler does not start any new tasks immediately after a task has been completed. The completion of a task causes a new one to be started only in the following situations:

• is called
• The task has been waiting some time on an order (the state running_waiting_for_order, when )
• The task could start but is delayed by the method
• The task is a process ( or ) and the process has not terminated itself immediately after starting

Limits the duration of a task operation (spooler_open, spooler_process etc.) or the whole task for a non-API task ( or ). An error does not occur when the priority of a job has not been set. Should a task exceed the time allowed, then the Job Scheduler aborts the task.

duration can be specified in seconds or in the HH:MM or HH:MM:SS formats.

Limits the idle time of an order controlled job (order="yes"). When a task is waiting on the next order and this idle time is exceeded, then the Job Scheduler ends the task.

The duration can be specified in seconds or in the HH:MM or HH:MM:SS formats.

idle_timeout="never" allows a job to run indefinitely, only limited by .

See also .

Note that this is only effective with and .

force_idle_timeout="yes" ends a task after idle_timeout has expired, even when fewer tasks than specified in min_tasks are running. min_tasks only starts new tasks after the task termination.

In this way tasks can be ended which may not take up a resource such as a database too long when idle.

Sets the priority of a task.

This attribute can be given the following values: idle, below_normal, normal, above_normal and high or the numerical values allowed for the operating system being used.

An error does not occur when the priority of a job is not set.

A task with a higher priority can block the computer.

See .

temporary="yes" defines a job as being temporary. This setting is only for . A job will then be deleted after being carried out and will no longer be recognized.

Is only effective when a job runs as its own process (see ) and either the job or monitor is implemented in Java. The options are handed over together with the Java options. The interpretation of these options depends on Java.

visible="no" and visible="never" make a job invisible in the results of and .

The Job Scheduler makes a job visible as soon as a task has been loaded when visible="no" is set. When visible="never" is set, then this job will never be returned when querying using the command.

Is only relevant for UNIX systems.

A job whose task process ends with a signal, causes the job to be stopped. Signals are sent when the task ends either by way of the kill system command or by way of a program being aborted.

If ignore_signals has not been specified, then a task ending with a signal stops the job (with the message ).

ignore_signals="all" means that a job will not be stopped by a signal.

A list of signal names (separated by blanks) can be specified instead of "all". The following signal names are recognized, depending on the operating system: SIGHUP, SIGINT, SIGQUIT, SIGILL, SIGTRAP, SIGABRT, SIGIOT, SIGBUS, SIGFPE, SIGKILL, SIGUSR1, SIGSEGV, SIGUSR2, SIGPIPE, SIGALRM, SIGTERM, SIGSTKFLT, SIGCHLD, SIGCONT, SIGSTOP, SIGTSTP, SIGTTIN, SIGTTOU, SIGURG, SIGXCPU, SIGXFSZ, SIGVTALRM, SIGPROF, SIGWINCH, SIGPOLL, SIGIO, SIGPWR und SIGSYS. Signal names which are not recognized by an operating system are ignored and a warning given.

Note that because a task ending with a signal which may be ignored can cause a TCP connection error (ECONNRESET), the Job Scheduler is so configured that TCP connection errors only lead to a job being stopped when ignore_signals="…" does not apply. The Job Scheduler reacts to this situation with the message.

See also and the Gnu man 7 signal command.

The default stop_on_error="yes" setting stops a job when a task ends with an exception or has an "[ERROR]" message in its log file. An error message can then be written, for example with .

Specifying stop_on_error="no" does not allow a job to be stopped in this situation. Should end with an exception, then the Job Scheduler sets the order in the error state ().

This setting is not effective, when either or are used.

replace="yes" causes any already existing job definition to be replaced. Should this not be possible - for example because a task is running - then the Job Scheduler will replace the job later.

An order controlled job () cannot be replaced.