activemq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Timothy Bish <tabish...@gmail.com>
Subject Re: [ActiveMQCPP 3.8.2|Linux] crash in activemq::core::ConnectionAudit::isDuplicate
Date Wed, 02 Jul 2014 10:20:21 GMT
On 07/02/2014 04:19 AM, Lyubov Romanchuk wrote:
> Ok. But this function doesn't only check for duplicates it changes the members as well.
> And it means that ConnectionAudit::isDuplicate must be thread safe because may be called
from different threads and as I see it's not.
>
>
> Software Developer | Check Point Software Technologies Ltd.
>
> ( Phone:     +972 73 2264585
> * Email:        lubar@checkpoint.com
> , Address:   Ha'solelim 5 St. Tel Aviv 67897, Israel
>
>
> -----Original Message-----
> From: Timothy Bish [mailto:tabish121@gmail.com]
> Sent: Tuesday, July 01, 2014 6:23 PM
> To: dev@activemq.apache.org
> Subject: Re: [ActiveMQCPP 3.8.2|Linux] crash in activemq::core::ConnectionAudit::isDuplicate
>
> On 07/01/2014 11:18 AM, Lyubov Romanchuk wrote:
>> But what is the purpose of checkForDuplicates? And which functionality will be disabled
in case it's false.
>>
> Surprisingly it checks for duplicates and filter them out in cases where your client
might receive a duplicate message from the broker such as failover disconnect / reconnect.
>
>> -----Original Message-----
>> From: Timothy Bish [mailto:tabish121@gmail.com]
>> Sent: Tuesday, July 01, 2014 6:08 PM
>> To: dev@activemq.apache.org
>> Subject: Re: [ActiveMQCPP 3.8.2|Linux] crash in
>> activemq::core::ConnectionAudit::isDuplicate
>>
>> On 07/01/2014 10:37 AM, Lyubov Romanchuk wrote:
>>> Hi,
>>>
>>> Thank you for the response.
>>> 1) It's very hard to support the test case for reproduction of this
>>> issue because  I don't know the root cause of the core and we have
>>> too complicated environment in order to try to simulate this
>>> behaviour
>>> 2) It seems like ConnectionAudit::isDuplicate is not thread safe but may be called
from different threads.
>>> 3) And I'm not sure that I understand correctly the purposes of isDuplicate.
Could you please explain what is it used for and if it's configurable?
>>>
>>> Best regards,
>>> Lyuba
>>>
>>> -----Original Message-----
>>> From: Timothy Bish [mailto:tabish121@gmail.com]
>>> Sent: Tuesday, July 01, 2014 12:20 AM
>>> To: dev@activemq.apache.org
>>> Subject: Re: [ActiveMQCPP 3.8.2|Linux] crash in
>>> activemq::core::ConnectionAudit::isDuplicate
>>>
>>> On 06/29/2014 10:16 AM, Lyubov Romanchuk wrote:
>>>> Hi,
>>>> We use activemqcpp library from the version 3.8.2.
>>>> Sometimes there is a crash in activemq::core::ConnectionAudit::isDuplicate.
>>>> Could you please advise with this core - what may cause such crash, is it
a known issue?
>>>>
>>>> Here is the call stack from one of the cores:
>>>>
>>>> * 1 Thread 32129  0x760417ab in
>>>> decaf::lang::PointerComparator<activemq::commands::ActiveMQDestinati
>>>> o n, decaf::util::concurrent::atomic::AtomicRefCounter>::operator()
>>>> (this=0x874ce1c, left=..., right=...) at decaf/lang/Pointer.h:243
>>>> (gdb) bt
>>>> #0  0x760417ab in
>>>> decaf::lang::PointerComparator<activemq::commands::ActiveMQDestinati
>>>> o n, decaf::util::concurrent::atomic::AtomicRefCounter>::operator()
>>>> (this=0x874ce1c, left=..., right=...) at decaf/lang/Pointer.h:243
>>>> #1  0x76108cad in
>>>> std::_Rb_tree<decaf::lang::Pointer<activemq::commands::ActiveMQDesti
>>>> n ation, decaf::util::concurrent::atomic::AtomicRefCounter>,
>>>> std::pair<decaf::lang::Pointer<activemq::commands::ActiveMQDestinati
>>>> o n, decaf::util::concurrent::atomic::AtomicRefCounter> const,
>>>> decaf::lang::Pointer<activemq::core::ActiveMQMessageAudit,
>>>> decaf::util::concurrent::atomic::AtomicRefCounter> >,
>>>> std::_Select1st<std::pair<decaf::lang::Pointer<activemq::commands::A
>>>> c tiveMQDestination,
>>>> decaf::util::concurrent::atomic::AtomicRefCounter>
>>>> const, decaf::lang::Pointer<activemq::core::ActiveMQMessageAudit,
>>>> decaf::util::concurrent::atomic::AtomicRefCounter> > >,
>>>> decaf::lang::PointerComparator<activemq::commands::ActiveMQDestinati
>>>> o n, decaf::util::concurrent::atomic::AtomicRefCounter>,
>>>> std::allocator<std::pair<decaf::lang::Pointer<activemq::commands::Ac
>>>> t iveMQDestination,
>>>> decaf::util::concurrent::atomic::AtomicRefCounter>
>>>> const, decaf::lang::Pointer<activemq::core::ActiveMQMessageAudit,
>>>> decaf::util::concurrent::atomic::AtomicRefCounter> > > >::find
>>>> (this=0x874ce14, __k=...) at
>>>> /packages/gnu-3.2.3-20.linux30/include/c++/3.2.3/bits/stl_tree.h:669
>>>> #2  0x76100316 in decaf::util::StlMap<decaf::lang::Pointer<activemq::commands::ActiveMQDestination,
decaf::util::concurrent::atomic::AtomicRefCounter>, decaf::lang::Pointer<activemq::core::ActiveMQMessageAudit,
decaf::util::concurrent::atomic::AtomicRefCounter>, decaf::lang::PointerComparator<activemq::commands::ActiveMQDestination,
decaf::util::concurrent::atomic::AtomicRefCounter> >::containsKey (this=0x874ce10, key=...)
>>>>         at
>>>> /packages/gnu-3.2.3-20.linux30/include/c++/3.2.3/bits/stl_map.h:345
>>>> #3  0x760fb57a in activemq::core::ConnectionAudit::isDuplicate
>>>> (this=0x874cdf4, dispatcher=0xc0a7e94, message=...) at
>>>> decaf/util/StlMap.h:794
>>>> #4  0x7603605c in activemq::core::ActiveMQConnection::isDuplicate
>>>> (this=0x8a74408, dispatcher=0x32332d46, message=...) at
>>>> decaf/util/concurrent/atomic/AtomicRefCounter.h:46
>>>> #5  0x76141bb5 in
>>>> activemq::core::kernels::ActiveMQConsumerKernel::dispatch
>>>> (this=0xc0a7e88, dispatch=...) at
>>>> activemq/core/kernels/ActiveMQSessionKernel.h:330
>>>> #6  0x760e4b1b in activemq::core::ActiveMQSessionExecutor::dispatch
>>>> (this=0x32332d46, dispatch=...) at decaf/lang/Pointer.h:258
>>>> #7  0x760e4e92 in activemq::core::ActiveMQSessionExecutor::iterate
>>>> (this=0x5a5676a0) at activemq/core/ActiveMQSessionExecutor.cpp:181
>>>> #8  0x762659b6 in activemq::threads::DedicatedTaskRunner::run
>>>> (this=0x5a6935e0) at activemq/threads/DedicatedTaskRunner.cpp:141
>>>> #9  0x7646285e in decaf::lang::Thread::run (this=0x32332d46) at
>>>> decaf/lang/Thread.cpp:141
>>>> #10 0x76421c17 in (anonymous namespace)::runCallback
>>>> (arg=0x5abb66e8) at decaf/internal/util/concurrent/Threading.cpp:266
>>>> #11 0x76421b89 in (anonymous namespace)::threadEntryMethod
>>>> (arg=0x5abb66e8) at decaf/internal/util/concurrent/Threading.cpp:254
>>>> #12 0x757f145b in start_thread () from /lib/libpthread.so.0
>>>> #13 0x7566024e in clone () from /lib/libc.so.6
>>>> (gdb) t 10
>>>> [Switching to thread 10 (Thread 32093)]#0  0x756951b2 in
>>>> dl_iterate_phdr () from /lib/libc.so.6
>>>> (gdb) bt
>>>> #0  0x756951b2 in dl_iterate_phdr () from /lib/libc.so.6
>>>> #1  0x0849cd55 in _Unwind_Find_FDE ()
>>>> #2  0x0849a815 in ?? ()
>>>> #3  0x0849b0a8 in _Unwind_RaiseException ()
>>>> #4  0x75762309 in __cxa_throw () from /usr/lib/libstdc++.so.5
>>>> #5  0x760a6bb8 in activemq::core::ActiveMQMessageAudit::isDuplicate
>>>> (this=0xbbef390, msgId=...) at decaf/util/HashMap.h:971
>>>> #6  0x760faba1 in activemq::core::ConnectionAudit::isDuplicate
>>>> (this=0x874cdf4, dispatcher=0x4ff9473c, message=...) at
>>>> decaf/lang/Pointer.h:258
>>>> #7  0x7603605c in activemq::core::ActiveMQConnection::isDuplicate
>>>> (this=0x8a74408, dispatcher=0xb726488, message=...) at
>>>> decaf/util/concurrent/atomic/AtomicRefCounter.h:46
>>>> #8  0x76141bb5 in
>>>> activemq::core::kernels::ActiveMQConsumerKernel::dispatch
>>>> (this=0x4ff94730, dispatch=...) at
>>>> activemq/core/kernels/ActiveMQSessionKernel.h:330
>>>> #9  0x760e4b1b in activemq::core::ActiveMQSessionExecutor::dispatch
>>>> (this=0xb726488, dispatch=...) at decaf/lang/Pointer.h:258
>>>> #10 0x760e4e92 in activemq::core::ActiveMQSessionExecutor::iterate
>>>> (this=0xb25f658) at activemq/core/ActiveMQSessionExecutor.cpp:181
>>>> #11 0x762659b6 in activemq::threads::DedicatedTaskRunner::run
>>>> (this=0xc073290) at activemq/threads/DedicatedTaskRunner.cpp:141
>>>> #12 0x7646285e in decaf::lang::Thread::run (this=0xb726488) at
>>>> decaf/lang/Thread.cpp:141
>>>> #13 0x76421c17 in (anonymous namespace)::runCallback (arg=0xb6ab2c8)
>>>> at decaf/internal/util/concurrent/Threading.cpp:266
>>>> #14 0x76421b89 in (anonymous namespace)::threadEntryMethod
>>>> (arg=0xb6ab2c8) at decaf/internal/util/concurrent/Threading.cpp:254
>>>> #15 0x757f145b in start_thread () from /lib/libpthread.so.0
>>>> #16 0x7566024e in clone () from /lib/libc.so.6
>>>>
>>>> Best regards,
>>>> Lyuba
>>>>
>>> No know issue in this area.  A test case to reproduce would be helpful.
>>>
>> You can set connection.checkForDuplicates=false on the URI or call
>> factory.setCheckForDuplicates(false) on your ActiveMQConnectionFactory instance to
disable duplicate checking which should sort you out.
>>
>> --
>> Tim Bish
>> Sr Software Engineer | RedHat Inc.
>> tim.bish@redhat.com | www.fusesource.com | www.redhat.com
>> skype: tabish121 | twitter: @tabish121
>> blog: http://timbish.blogspot.com/
>>
>>
>>
>> Email secured by Check Point
>>
>
> --
> Tim Bish
> Sr Software Engineer | RedHat Inc.
> tim.bish@redhat.com | www.fusesource.com | www.redhat.com
> skype: tabish121 | twitter: @tabish121
> blog: http://timbish.blogspot.com/
>
>
> Email secured by Check Point
>
We welcome contributions so if you think you've found an issue go ahead 
and dive in and see if you can work up a patch and test to cover the 
scenario.

-- 
Tim Bish
Sr Software Engineer | RedHat Inc.
tim.bish@redhat.com | www.fusesource.com | www.redhat.com
skype: tabish121 | twitter: @tabish121
blog: http://timbish.blogspot.com/


Mime
View raw message