Hi Mike, Thanks for the tips. I've debugged through it and it turns out that the stop() method of DefaultRMIHost doesn't even get called. I've done some local changes that call that stop method (even though it's really ugly as I had to use casting because the RMIHost interface doesn't have a stop method and ExtensibleRMIHost doesn't provide a stop method through it's public API). Even with those hacks, the problem is still reproducible... I'm trying to figure out a way of monitoring what's happening in the concurrent environment as simple debugging isn't that helpful in reproducing it.
Florian On Wed, Feb 9, 2011 at 12:08 PM, Mike Edwards < mike.edwards.inglen...@gmail.com> wrote: > On 07/02/2011 10:56, Florian Moga wrote: > >> On Mon, Feb 7, 2011 at 12:43 PM, Simon Laws <simonsl...@googlemail.com >> <mailto:simonsl...@googlemail.com>> wrote: >> >> So when you do a full build the build always hangs? >> >> Yes. >> >> Is that test still in the build as I don't experience that. >> >> Yes. >> >> What environment are you on? >> >> Ubuntu 10.10, Sun JVM 6u22, Maven 2.2.1, Ant 1.7.1 >> Same happens with OpenJDK >> >> >> Simon >> >> >> -- >> Apache Tuscany committer: tuscany.apache.org < >> http://tuscany.apache.org> >> Co-author of a book about Tuscany and SCA: tuscanyinaction.com < >> http://tuscanyinaction.com> >> >> >> Florian, > > My experience with this testcase is that it fails with a hang on an > intermittent basis, relatively infrequently. > > I am running on Windows XP with Sun JDK 1.6.0_22. > > I looked into reports of hangs relating to the RMI code and there are > reports of such hangs if the RMI Registry is not released correctly when > stopping it. RMI uses a pile of threads and can have connections open > between Client and Server side that need to be properly closed. > > A thread dump when the hang occurs does show an RMI thread waiting on some > kind of lock, but I am not able to make much sense of what is going on. > > I suspect that the place to start looking is in the RMIHost code - and in > the stop() method of DefaultRMIHost... > > > Yours, Mike. >
