[
https://issues.apache.org/jira/browse/QPID-2214?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12785428#action_12785428
]
Andrew Stitcher commented on QPID-2214:
---------------------------------------
The "correct" way to fix this is to ensure that you never create an unbounded
number of io threads.
The design intention was always that the number of io threads was small and
fixed, and the client code breaks that intention.
I'm looking at a fix that stops the client code from creating and then
destroying a thread per connection.
> memory leak in qpid::client::Connection
> ---------------------------------------
>
> Key: QPID-2214
> URL: https://issues.apache.org/jira/browse/QPID-2214
> Project: Qpid
> Issue Type: Bug
> Components: C++ Client
> Affects Versions: 0.5
> Environment: qpid 0.5 on a Debian Linux with gcc 4.3.3.
> Reporter: Daniel Etzold
> Assignee: Andrew Stitcher
> Priority: Critical
> Attachments: qpid-memleak.patch
>
>
> Hi,
> when executing the code below (connecting and disconnecting to a local broker
> without sending any messages) the memory usage increases constantly and
> rapidly. After 10.000 iterations several hundred megabytes of resident memory
> are used.
> When commenting out the lines "connection.open()" and "close()" the memory
> usage does not increase.
> So, is there a memory leak in Connection open/close?
> int
> main(int argc, char** argv)
> {
> while (1) {
> qpid::client::Connection connection;
> connection.open("localhost", 5672);
> connection.close();
> }
> }
> Running my test binary with valgrind (the loop is limited to 100 iterations)
> valgrind --leak-check=full ./myqpidtest
> it seems that valgrind does not find any memory leaks.
> =17321== 76 bytes in 1 blocks are definitely lost in loss record 2 of 2
> ==17321== at 0x4007ADE: calloc (vg_replace_malloc.c:279)
> ==17321== by 0x430CD4E7: (within /lib/ld-2.3.6.so)
> ==17321== by 0x430CD58B: _dl_allocate_tls (in /lib/ld-2.3.6.so)
> ==17321== by 0x43AE428F: pthread_create@@GLIBC_2.1 (in
> /lib/tls/i686/cmov/libpthread-2.3.6.so)
> ==17321== by 0x43AE4AA7: pthread_cre...@glibc_2.0 (in
> /lib/tls/i686/cmov/libpthread-2.3.6.so)
> ==17321== by 0x4AA3B55:
> qpid::sys::ThreadPrivate::ThreadPrivate(qpid::sys::Runnable*) (in
> libqpidcommon.so.0.1.0)
> ==17321== by 0x4AA3682: qpid::sys::Thread::Thread(qpid::sys::Runnable*)
> (in libqpidcommon.so.0.1.0)
> ==17321== by 0x4C8BC6B: qpid::client::TCPConnector::init() (in
> libqpidclient.so.0.1.0)
> ==17321== by 0x4C8040A: qpid::client::ConnectionImpl::open() (in
> libqpidclient.so.0.1.0)
> ==17321== by 0x4C6CD4A:
> qpid::client::Connection::open(qpid::client::ConnectionSettings const&) (in
> libqpidclient.so.0.1.0)
> ==17321== by 0x4C6CED2: qpid::client::Connection::open(std::string const&,
> int, std::string const&, std::string const&, std::string const&, unsigned
> short) (in libqpidclient.so.0.1.0)
> ==17321== by 0x805EE1E: main (myqpidtest.cpp:281)
> ==17321==
> ==17321== LEAK SUMMARY:
> ==17321== definitely lost: 76 bytes in 1 blocks.
> ==17321== possibly lost: 0 bytes in 0 blocks.
> ==17321== still reachable: 48 bytes in 3 blocks.
> ==17321== suppressed: 0 bytes in 0 blocks.
> The 76 bytes which are reported to be definitely lost is constant whether
> running the loop 100 time or 1000 times.
> Regards,
> Daniel
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
Apache Qpid - AMQP Messaging Implementation
Project: http://qpid.apache.org
Use/Interact: mailto:dev-subscr...@qpid.apache.org
