Sorry for late response,I've created this github repo that you can use to test
this: RusuIonutMihai/TestArtemisMessageExpiration (github.com). The details are
in this README file.
Also I want to mention other things that I've noticed: After the problem
occurs, if the broker is restarted, then the messages appear in the expiry
queue and the problem seem to not happen again. To make the problem appear
again, delete the queue that receives the messages (Q1) and restart the broker.
Seems strange, but this is what I've encounter in my tests.
Pe marți, 12 martie 2024 la 16:38:52 EET, Justin Bertram
<jbert...@apache.org> a scris:
I'm not aware of any problems like this.
Can you provide a way to reproduce it?
Justin
On Tue, Mar 12, 2024 at 9:27 AM rusu ionut <ionut_rusu_2...@yahoo.ro.invalid>
wrote:
> Hello,I'm using latest version of Artemis with AmqpNetLite from .NET to
> send/receive messages. I have the following scenario: one queue (Q1) with
> an expiry address set. Q1 will receive messages with Ttl of 1 minute. So
> messages will be available in Q1 for 1 minute and then are moved to expiry
> address/queue. I have a receiver set for the expiry queue that save the
> messages to a DB. For this test, I don't have any receiver for Q1, so all
> the messages should be saved in my DB eventually. I'm sending 2000
> messages/second for 10 minutes. The problem is that sometimes some messages
> are still shown in the queue attributes (message count, persistent size,
> ...). The attribute Messages expired is the same number as the number of
> messages that I have in the DB. If I go to Browse Queue, no message is
> shown. Also if I connect a receiver to Q1 I don't receive any of the lost
> messages, but if I send new messages, I receive the new ones. I also tried
> Purge queue and nothing was changed. So message count is not 0, but the
> messages could not be accessed. All the messages are durable and have the
> same size = 282 bytes. The only parameter that I changed regarding
> expiration was message-expiry-scan-period = 5000.Is this a known problem?
> Thank you in advance!
