An order is the instruction set describing how a job chain should execute a series of jobs. An order can be carried out immediately or when a preconfigured event takes place. Example events are times of the day, days of the week or files being added to a directory.

A job chain can be seen as an assembly line on which orders are run. The orders are processed by the individual machines of the assembly line one after the other. A job comprises exactly one step in the processing of an order. After a job has been completed, the state of the order is changed, which in turn affects the next step in the job chain. (For example, after completing one job, an order could be added to the queue for the next job.) This process will be repeated until the order reaches the end of the job chain.

An order has an identifier, which is valid within that job chain, and a readable title. The order also has a status, which changes after the processing of each job. An order can contain parameters, which are handed on to all the jobs in the job chain. It can also carry a load - i.e. an individual XML document, which it makes available to the jobs.

Orders allow jobs and job chains to be reused: an order can be allocated to a single job chain and define the time at which the job chain starts execution. This configuration is often used, when each job is only used in a single job chain and the parameters are set for each job.

It is also possible to configure more than one order for the same job chain. In this case, the orders can be given different parameters, which are then handed over to the jobs, so that the same jobs can be used for different purposes.

See the class.

Orders can also be transferred using TCP and the command.

Further, the Job Scheduler HTML interface provides operations with which orders can be manually started.

Job Chains

Job chains define a series of jobs, whose execution is activated by an order. Job chains specify job dependencies for the successful and incorrect execution of the jobs. They can be restarted - that is, the Job Scheduler saves the state of an order within a job chain in a database. Should the processing of an order be broken off at a particular job in a job chain, on restarting, the Job Scheduler is able to restore the order at the job within the job chain, at which processing was broken off.

Job chains are described in the and classes.

A job is made order controlled using the setting.

In order that a job can be used in a job chain, it should be defined using . If this is not done, the jobs can be separately called up as standalone jobs.

Order Queue and Tasks

Every job which is order controlled has an order queue in which orders which are to be processed are collected. The Job Scheduler starts processing an order as soon as it is added to the queue and the element allows it to start. Should there be a number of orders in the job queue and the job allows multiple tasks, (), then the Job Scheduler will process the tasks in parallel.

The Job Scheduler hands an order over to a task by storing the order in the method and calling up the method. The spooler_process() method ends with either a true or false result, which, in turn, determines whether or not an order is processed by the next job in the job chain (see ).

Recognition of Double Orders

When an order is added to the job queue, the Job Scheduler checks whether an order with the same id has already been entered (only, however, when the order id has been set). Should the order queue or job chain already have an order with the same id, then the already existing order will be overwritten. Should the new order have a different priority to the original, then the scheduler will take this into account.

Directory Monitoring with File Orders

See for information about orders which are created from the files in a directory.

Priority

Order priority is set using . Orders with a higher priority are put by the Job Scheduler at the front of the queue.

Ending Tasks

The Job Scheduler lets tasks exist as long as the parameter is valid. They then have the status running_waiting_for_order.

The maximum time which the Job Scheduler will wait before a task is automatically ended is set using . This time should be set to a value such as 30 sec., so that resources can be set free.

Should the Job Scheduler require to start a task for which no more process class resources are available, then it automatically looks for another task of the same process class with the running_waiting_for_order status, which it then terminates.

Tasks can be terminated using , and .

A task ends when is implemented. The order is then given the successor state.

Accelerated Order Processing

Jobs which are order controlled are prioritized. The element contains an arbitrary value which defines the position of orders of the same status in the order queue. This value must not exceed the limit defined by the attribute.

The Job Scheduler gives orders a higher priority the further advanced they are in the job chain. This helps reduce congestion and means that jobs which are started first will be more quickly completed. It is as though the machines at the end of an assembly line were running more quickly than those at the beginning.

Jobs which are order controlled have, in general, priority over jobs which are not order controlled.

Database

More information about the longer term storage of order queues and the order history is to be found in the section on databases.