Hi Martin, ah ok, this should be the reason. So, if I slow down the send I shouldn't see the memory growing so fast, is that correct ?
On Mar 30, 2010, at 5:47 PM, Martin Sustrik wrote: > Hi David, > >> I think I found a memory leak. You can reproduce the problem just >> running the local_thr and remote_thr executable under perf. >> The memory leak is associated to the sending of messages, so just >> looking at the process running the remote_thr you can see that the >> memory grows indefinitely. >> Any clue ? > > Can you check with valgrind? It reports no leaks on my box. > > As for the growing memory usage it may be caused by sender being faster > than receiver. In such a case messages are stored in the in-memory pipe > till they are consumed by the receiver. > > On most operating systems the memory allocated by malloc is returned to > the OS only when process exits. That may be the reason you don't see > memory usage dropping after the test is over. > > Martin > _______________________________________________ > zeromq-dev mailing list > [email protected] > http://lists.zeromq.org/mailman/listinfo/zeromq-dev _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
