| Author |
Message
|
| basa_bbc |
 Posted: Tue Jan 04, 2005 1:28 am Post subject: API EXIT |
 |
|
Apprentice
Joined: 20 Jan 2004
Posts: 35
|
Hello,
We ahve developed an API exit to perform certain operations on our production quue manager. We have 200 Queue objects in the Queue Manager, But when ever an event happens on any of the queue on the queue manager the exit is invoked.
Our requirement is to control the API exit invocation only for certain queues. One way of doing this is by putting the queue names in the namelist and accessing the namelist thorugh API exit.
I am really interested in knowing if there are any configurations that can be set at EXIT level for the same.
Thanks for the speedy reply.
Regards,
bbc |
|
| Back to top |
|
 |
| jefflowrey |
| Posted: Tue Jan 04, 2005 5:20 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
The API exit is a property of the queue manager, not a specific queue.
So, no, there is no mechanism out of the box to specify certain queues for the API exit to be run for, and not others.
But the namelist idea is good.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| basa_bbc |
| Posted: Tue Jan 04, 2005 5:33 am Post subject: API Exit |
|
|
Apprentice
Joined: 20 Jan 2004
Posts: 35
|
Jefflowery,
Thanks for the clarification. I am little confused, can you tell me that if i have two API exits running on the same queue manager and if they are event based(on the queue) , then both the API exits will consume lot of memory , because if an event gets generated on any queue object not related to that particular API exit , the API exit will be initiated.
This is a waste of resource, can you please help me in another for of solution or is this the limit of API exits.
BBc |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Tue Jan 04, 2005 5:37 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
I think it is a limitation of the API exit.
Exactly as you have noticed, API exits get invoked for every single API call. This means that they can have a large effect on performance, if care is not take to write them properly.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| basa_bbc |
| Posted: Tue Jan 04, 2005 5:48 am Post subject: API Exit |
|
|
Apprentice
Joined: 20 Jan 2004
Posts: 35
|
Jefflowery,
Thanks for confirmation. Could you provide me some links wherein i can have complete information regarding , points to be considered while coding for API Exists.
Regards,
bbc |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Tue Jan 04, 2005 6:02 am Post subject: Re: API Exit |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
| basa_bbc wrote: |
| Could you provide me some links wherein i can have complete information regarding , points to be considered while coding for API Exists. |
The button at the top of the page, labeled "Documentation".
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| clindsey |
| Posted: Tue Jan 04, 2005 7:48 am Post subject: |
|
|
Knight
Joined: 12 Jul 2002
Posts: 586
Location: Dallas, Tx
|
BBC,
I prefer to read the namelist at startup of the process and cache the information. It is a much faster lookup. Go to http://www.developer.ibm.com/tech/sampmq.html and search for mirrorq. The source code is included and it will show you a good way to do this. If you have exits that are chained, each can have it's own namelist and each will manage it's own cache. Also keep in mind that you only need to register the APIs that you generate an event for. For example, if you have no need to process and event for MQINQ then do not register this function in the exit.
Hope this helps,
Charlie |
|
| Back to top |
|
|
| Vijay Kumar P |
| Posted: Wed Feb 16, 2005 10:10 pm Post subject: No Automated / Builtin feature. |
|
|
Novice
Joined: 24 Jan 2005
Posts: 10
|
Hello Balaji,
Jef and Clind are right. Adding to there points.
We need to explicitly handle the execution of code for the specific queue using some of the mutex, critical sections or some sync mechanism to solve the problem.
I believe some R&D on advanced shared object coding will help.
Thanks & Regards,
Vijay. |
|
| Back to top |
|
|
|
|
