I sent a response about it on github of what I think is going on:
https://github.com/apache/activemq/pull/1641#issuecomment-3823774532

I think we just need a very short sleep for now unless we figure out
another way in the Advisory broker without deadlocking

On Fri, Jan 30, 2026 at 4:34 AM Jean-Louis Monteiro <
[email protected]> wrote:

> I am not happy at all with this fix. We should release the lock before any
> write operation and fire is a write operation under the cover. So this will
> create a deadlock sooner or later.
> I think I was wrong thinking it's a production code issue. It is a race
> condition issue in the test I believe.
>
> My apologies for sending this maybe too early. Still appreciate your
> thoughts, please.
> --
> Jean-Louis Monteiro
> http://twitter.com/jlouismonteiro
> http://www.tomitribe.com
>
>
> On Fri, Jan 30, 2026 at 9:50 AM Jean-Louis Monteiro <
> [email protected]> wrote:
>
> > Hi,
> >
> > I noticed that AdvisoryBrokerTest is randomly failing.
> > For example, here is the error
> >
> > 2026-01-27T14:10:54.4229847Z [ERROR]
>  
> AdvisoryBrokerTest>CombinationTestSupport.runBare:107->testConsumerAdvisoriesReplayed:175
> ClassCast class org.apache.activemq.command.ConsumerInfo cannot be cast to
> class org.apache.activemq.command.RemoveInfo
> (org.apache.activemq.command.ConsumerInfo and
> org.apache.activemq.command.RemoveInfo are in unnamed module of loader
> 'app')
> >
> >
> > I dug into it and ended up thinking it's a bug in AdvisoryBroker as
> > opposed to the test itself.
> > I created https://issues.apache.org/jira/browse/AMQ-9852
> > And the PR https://github.com/apache/activemq/pull/1641
> >
> > I'd appreciate a review in case my understanding is incorrect.
> >
> > --
> > Jean-Louis Monteiro
> > http://twitter.com/jlouismonteiro
> > http://www.tomitribe.com
> >
>

Reply via email to