[
https://issues.apache.org/activemq/browse/AMQCPP-90?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Albert Strasheim reopened AMQCPP-90:
------------------------------------
Revision 519590 seems to have broken our library's one unit test.
The test does the following:
- creates a connection
- sets an exception listener on the connection
- creates an asynchronous session
- uses the session to create one producer and one consumer associated with a
topic
- sets a message listener on the consumer
- starts the connection
The test then uses the producer to send a BytesMessage to the consumer. The
BytesMessage contains a single integer. The message listener tries to read two
integers from the BytesMessage.
In revision 519589 this triggered the exception listener, which signals our
test method through a latch and the test completes normally.
With revision 519590, reading the second integer causes the test to print:
Reached the end of the buffer
FILE: ..\src\main\activemq\io\ByteArrayInputStream.cpp, LINE: 104
FILE: ..\src\main\activemq\io\DataInputStream.cpp, LINE: 314
FILE: ..\src\main\activemq\io\DataInputStream.cpp, LINE: 104
FILE: ..\src\main\activemq\io\DataInputStream.cpp, LINE: 181
FILE: ..\src\main\activemq\core\ActiveMQConsumer.cpp, LINE: 378
to the console and then the program hangs. We also see this behaviour with the
latest HEAD revision (519654).
I'm also seeing an intermittent segfault (can't reproduce it reliably yet). The
stack trace looks like this:
>
> unittests.exe!activemq::core::ActiveMQConsumer::dispatch(activemq::core::DispatchData
> & data={...}) Line 368 + 0x16 bytes C++
unittests.exe!activemq::core::ActiveMQSessionExecutor::dispatch(activemq::core::DispatchData
& data={...}) Line 140 C++
unittests.exe!activemq::core::ActiveMQSessionExecutor::dispatchAll()
Line 185 C++
unittests.exe!activemq::core::ActiveMQSessionExecutor::run() Line 152
C++
unittests.exe!activemq::concurrent::Thread::runCallback(void *
param=0x008d3748) Line 152 + 0x13 bytes C++
msvcr80d.dll!102047c1()
[Frames below may be incorrect and/or missing, no symbols loaded for
msvcr80d.dll]
msvcr80d.dll!10204767()
kernel32.dll!7c80b683()
This seems to also have caused by a recent change, as I've never seen this
before today.
> Openwire Session doesn't run its MessageListeners from a single thread
> ----------------------------------------------------------------------
>
> Key: AMQCPP-90
> URL: https://issues.apache.org/activemq/browse/AMQCPP-90
> Project: ActiveMQ C++ Client
> Issue Type: Bug
> Components: Openwire
> Affects Versions: 2.0
> Reporter: Albert Strasheim
> Assigned To: Nathan Mittler
> Fix For: 2.0
>
> Attachments: MultipleConsumerTest.cpp, MultipleConsumerTest.java
>
>
> According to the JMS 1.1 specification, section 4.4.14 Serial Execution of
> Client Code:
> For this reason, JMS does not cause concurrent execution of client code unless
> a client explicitly requests it. One way this is done is to define that a
> session
> serializes all asynchronous delivery of messages.
> To receive messages asynchronously, a client registers an object that
> implements the JMS MessageListener interface with a MessageConsumer. In
> effect,
> a Session uses a single thread to run all its MessageListeners. While the
> thread is
> busy executing one listener, all other messages to be asynchronously delivered
> to the session must wait.
> In the attached code I have 5 consumers created from the same session, all
> with the same message listener instance that receives a message, prints
> something and then sleeps for a while. When running the C++ code, one sees
> that 5 messages (there are 5 consumers) are delivered at approximately the
> same time. Under Java, one only sees 1 message delivered per interval.
> This seems to indicate that the Openwire Session doesn't run its
> MessageListeners from a single thread. This can cause problems if the user's
> MessageListener isn't thread safe.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
