Ab Initio queues are an adaptation of the first-in/first-out queue concept:
• They provide record-based persistence.
• Publishers write data to the queue.
• Subscribers to the queue read the data in the order it was written.
In many ways, Ab Initio queues are analogous to multifiles in ordinary Ab Initio graphs. They provide a method for storing records in an ordered sequence of files. However, they also hold additional information necessary to allow both the removal of data from disk when it is no longer needed, and the recovery of graphs that stop running for any reason.
Queues are the most reliable method for storing continuous flow data. We recommend that you get your data into this format as early as possible in the data processing stream
You use the m_queue command to create an Ab Initio queue. When you execute this command, you specify the name of the directory or multidirectory that you want to contain the new queue. This is the value of the queue_path argument to the m_queue command. The Co>Operating System creates the directory you specify and sets up a queue inside it. A directory or multidirectory can contain only one queue. Consequently, although the directory is not the queue, the name of the directory that contains the queue identifies the queue.
Each Ab Initio queue directory contains a number of files, which contain the queue data and the queue infrastructure. If you look at the queue in the file system, you can see these files. However, you never directly operate on them.
m_queue command operartion:
m_queue create -f
Example
m_queue create -f /u05/cromwell/puranika/Projects/Cromwell/Registartion/data/input/queue sub1 sub2
Stopping Continuous Flow Graphs
Use one of the following two commands to stop a continuous flow graph:
• m_shutdown [-f | -status] job_name Waits until the graph commits the next checkpoint in order to end the job cleanly, then stops the execution of the graph and deletes all checkpoint temporary files.
• m_kill job_name The execution of the graph stops immediately
