When there is buildup on a queue, it usually means something needs to be repaired. Usually the most important information about queues is the destination, or “target,” of any message that has been too long on a queue. In general, when a queued message is not being sent, it is because it cannot get to its target. If you can find out what is causing a problem with the target, when you solve that problem, the queue buildup will generally disappear. For example:
-
For a business service or business operation, if a queue is suddenly longer, this generally means that there is a problem communicating with an external system. An external connection may be down, or there may be a peak-hour effect that is affecting throughput on your external connections.
-
For a business service or business operation, if a queue is consistently long, this generally means that there is a consistent delay in sending messages. You should probably examine the external connection to see if there a performance problem that you can solve. If that is not possible, you could increase the appropriate pool size (unless you need to ensure first-in-first-out processing).
See Pool Size and Actor Pool Size.
-
For a business process that uses a private pool, if a queue is consistently long, you could increase the appropriate pool size (unless you need to ensure first-in-first-out processing).
-
If the actor queue is suddenly longer, a business process may have experienced an error that has caused it to become “stuck” in some way.
-
If the actor queue is consistently long, the actor pool for the production may need to be larger.
-
If many queues have a consistently large buildup, there may be a general capacity issue on the host computer, the production (in its role as a CSP application) may need more resources, or the underlying InterSystems IRIS installation may need to be tuned. For suggestions, see the System Administration Guide. In general, however, you will be able to keep queues moving with the simpler adjustments listed in this topic.
