1752803
Members
5437
Online
108789
Solutions
SOLVED
The generic queue was created to handle either banks of similar devices, such as a row of printers, or a set of clustered systems.
In the case of printers, an execution queue (and symbiont process) is created for each, then a generic queue is created consisting of the execution queues. When a job comes in, it is sent to the first available execution queue in the generic queue's list. If jobs come in infrequently, the first printer in the list winds up doing most of the work. Some sites would have a batch job that would rotate the order of the execution queues periodically, so that each printer would have the chance of being the "primary" printer.
In the case of batch queues, an execution queue is created for each, then a generic queue is created consisting of the execution queues. However, with batch queues, there is no symbiont process and no inherent restriction of one active job per queue; depending on the characteristics of the system (processor speed, number of processors, amount of memory, available CPU after priority tasks and interactive users are handled, etc), some systems can easily handle many jobs simultaneously. Regardless, by setting the /JOB_LIMIT, we can control how many will run on a given host.
When a generic queue comes into play, the queue manager looks at the total number of job slots across all the (started) execution queues, then looks at the percentage of job slots contributed by a given system. It then sends the job from the generic queue to the execution queue with the most available slots relative to the overall slot percentage at the time the batch job starts execution. (What Fekko Stubbe observed.) Of course, sending a batch job to a particular execution queue will force the job to that queue's system, where it simply becomes just another job in the queue manager's subsequent queue loading calculations.
So by adjusting the /JOB_LIMIT and the order of the execution queues, you should be able to get the balance you desire.
Note: While I am an HPE Employee, all of my comments (whether noted or not), are my own and are not any official representation of the company
In the case of printers, an execution queue (and symbiont process) is created for each, then a generic queue is created consisting of the execution queues. When a job comes in, it is sent to the first available execution queue in the generic queue's list. If jobs come in infrequently, the first printer in the list winds up doing most of the work. Some sites would have a batch job that would rotate the order of the execution queues periodically, so that each printer would have the chance of being the "primary" printer.
In the case of batch queues, an execution queue is created for each, then a generic queue is created consisting of the execution queues. However, with batch queues, there is no symbiont process and no inherent restriction of one active job per queue; depending on the characteristics of the system (processor speed, number of processors, amount of memory, available CPU after priority tasks and interactive users are handled, etc), some systems can easily handle many jobs simultaneously. Regardless, by setting the /JOB_LIMIT, we can control how many will run on a given host.
When a generic queue comes into play, the queue manager looks at the total number of job slots across all the (started) execution queues, then looks at the percentage of job slots contributed by a given system. It then sends the job from the generic queue to the execution queue with the most available slots relative to the overall slot percentage at the time the batch job starts execution. (What Fekko Stubbe observed.) Of course, sending a batch job to a particular execution queue will force the job to that queue's system, where it simply becomes just another job in the queue manager's subsequent queue loading calculations.
So by adjusting the /JOB_LIMIT and the order of the execution queues, you should be able to get the balance you desire.
Note: While I am an HPE Employee, all of my comments (whether noted or not), are my own and are not any official representation of the company
