gtully commented on pull request #3858:
URL: https://github.com/apache/activemq-artemis/pull/3858#issuecomment-975372999
The new test fails for me, checking for messageCount on server node 1, b/c
the redistribution has already happened! Adding a small redistribution delay
sorts that, but it is a little flakey.
+++
b/tests/integration-tests/src/test/java/org/apache/activemq/artemis/tests/integration/cluster/distribution/MessageRedistributionTest.java
@@ -800,7 +800,7 @@ public class MessageRedistributionTest extends
ClusterTestBase {
String address = "test.address";
String queue = "test.address";
String clusterAddress = "test";
- AddressSettings settings = new
AddressSettings().setRedistributionDelay(0).setAutoCreateAddresses(true).setAutoCreateQueues(true);
+ AddressSettings settings = new
AddressSettings().setRedistributionDelay(100).setAutoCreateAddresses(true).setAutoCreateQueues(true);
Also, it makes me think that adding a redistributor initially will compete
in the normal case, where a binding is created before a consumer, if there is
no delay, then the consumer may not get any messages. But what should that
delay be? Does that need to be configurable? I think it does need to be > 0.
In a typical JMS case, with auto create queue, there is a tiny delay before
the consumer is created, so we really must have a delay or the normal
expectation will fail.
I guess the point is we are changing/extending the trigger for
redistribution, to be something added (a binding) rather than something removed
(a consumer).
Maybe in this new added case, we respect the redistribution delay if
configured and fallback to 5s if there is none, such that we don't break things.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
