ARTEMIS-4476 Possible isolated consumer with OpenWire #4694
Conversation
clebertsuconic
force-pushed
the
cleanup-openwire
branch
from
1aafc06
to
c4fd8a3
Compare
| Assert.assertTrue(beforeCreateCalled.await(5, TimeUnit.MINUTES)); | ||
|
|
||
| server.getRemotingService().getConnections().forEach(r -> { | ||
| r.fail(new ActiveMQException("this is a simulation")); |
There was a problem hiding this comment.
this would block with the current version as this is calling shutdown.
I'm blocking the thread on purpose there.
I could not reproduce the leakage here. But I believe the change I made would be better in this case.
I'm open to suggestions.
| } | ||
|
|
||
| if (executor != null) { | ||
| executor.execute(() -> doFail(me, message)); |
There was a problem hiding this comment.
@gtully notice that the executor is used by the Actor. Meaning it will always execute after sthudown is called, using the same thread it would have used with the actor.
This is just a safer option.
There was a problem hiding this comment.
I don't follow, the point is to terminate processing of commands and execute the doFail as the last/next task. The only call to fail should be from the netty socket handler that sees a socket error, remote close etc. It is the transport initiating a close on a socket error.
|
replaced by #4700 |
No description provided.