It looks like the Dispatcher was for data change notifications. I suspect a listener was hung or responding slowly so the actor's mailbox filled up with change notifications. I would suggest getting a thread dump next time.
On Mon, May 29, 2017 at 7:52 AM, Michael Vorburger <vorbur...@redhat.com> wrote: > Hi guys, > > I just ran MAT([1]) over an HPROF heap dump on OOM in Bug 7370, and it > (MAT) raises a "leak suspect" in akka.dispatch.Dispatcher - see the [3] > java_pid19570_Leak_Suspects.zip just attached to Bug 7370 ... questions: > > Is this perhaps something you jump at with an "ah that, we know about it > and already fixed that in ..." ? > > If not, how do we go about better understanding the root cause of this, > and be able to eventually fix this? > > My underlying assumption here is that isn't "normal" and not just "by > design" - if it is, I'd love some education... like I'm hoping that the > conclusion here isn't simply that MD SAL's data store is a dumb in-memory > data base which basically just takes a huge amount of GBs to keep (all) > YANG model instances on the heap - or is it? > > Tx, > M. > > [1] https://www.eclipse.org/mat/ > > [2] https://bugs.opendaylight.org/show_bug.cgi?id=7370 > > [3] https://bugs.opendaylight.org/attachment.cgi?id=1816 > > -- > Michael Vorburger, Red Hat > vorbur...@redhat.com | IRC: vorburger @freenode | ~ = http://vorburger.ch > > _______________________________________________ > mdsal-dev mailing list > mdsal-...@lists.opendaylight.org > https://lists.opendaylight.org/mailman/listinfo/mdsal-dev > >
_______________________________________________ controller-dev mailing list controller-dev@lists.opendaylight.org https://lists.opendaylight.org/mailman/listinfo/controller-dev
