|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
 |
|
Queue Message Expiration |
« View previous topic :: View next topic » |
Author |
Message
|
Cluniac |
Posted: Thu Jun 05, 2008 5:57 am Post subject: Queue Message Expiration |
|
|
Newbie
Joined: 05 Jun 2008 Posts: 1
|
Hello,
Recently our client has been introducing some linux based platforms into their webfarm. We did some simple MQ tests with the exact same deployed application running on AIX 5.3, and on Redhat Linux 4.4. On both environments we are running MQ 6.0.2.
For one of the tests, while the applications had simulated load on them placing messages on queues, we shut off the sender channel backend and kept the local queue managers running.
On the AIX platform, we noticed that the queue kept growing in size for 2 or 3 minutes. This makes sense, as the messsages are received from the application but have no where to go since the backend is shutoff. We did the exact same test on the Linux platform, but here we noticed a difference: After just 30 seconds or so, messages began expiring and the queue size was not growing at all.
We have investigated this, and so far have come up with the following:
We know that the application sets an expiration time of 20 seconds for every message being put on the queue. But we noticed in AIX, they were lasting much longer, seemingly never expiring, and in Linux they were expiring but not this fast.
The expired messages are normally only deleted after a MQGET.
When no MQGET is done the expired messages will stay in the queue (queueing) . Since Mq v6 , the qmanagers scans all the queues periodically for expired messages (so that the expired messages are deleted) .
The interval time between 2 scans is probably not fixed and it's not a setting you can change. (except on z/os).
The behaviour seen on aix tells us that the scan interval time is higher then on linux.
On Linux, because the system was performing higher we did the test with many more concurrent users, and a higher load on MQ. Perhaps its possible that the higher load on MQ causes this scan interval to be lower, and hence the difference that we saw.
Can anyone confirm our theories, and whether they have seen any similar behavior like this before?
Regards,
Matt |
|
Back to top |
|
 |
|
|
 |
|
Page 1 of 1 |
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|
|
|