Hello Carsten, from what I know there are 2 scenarions in the CAVE system of our customer when using multicast. Sometimes the systems starts und runs but is really slow, like it is when using streamsock communication. Sometimes it freezes/crashes during start up/initial synchronization phase.
We haven't tried to set multicast port and group options so far. - What would be the advantages? - Do we need to apply those only to all servers or to the client as well? - I'm not so familiar with multicast communication. Do we need to setup some options in the switches/routers of the custumer network other that enabling multicast? Regarding gdb: Unfortunately it is the installation of our custumer which is miles away. So gdb debugging is currently no option. Thanks, Michael Am 10.08.2011 15:52, schrieb Carsten Neumann: > Hello Hanno, > > On 08/10/2011 04:26 AM, Hanno Hugenberg wrote: >> Finaly we can send you some more informations. >> >> It seems, that the client freezes during waiting for the servers. >> >> This may happen, because we get a Fatal error on one of the linux servers. >> The error line is: >> >> "FATAL: Error in dgram reader thread:SocketLib: recvFrom() 14 Bad address" > yes, a crashing server would probably freeze the client, waiting for it > to acknowledge received packets. > Michael's original message indicated the system did run for a bit before > freezing, but from the above "Bad address" error it seems more like it > never really runs on all servers? > >> The server is started with the following code (snippets): >> >> connectionType = "Multicast"; >> address = ""; >> >> code: >> ----------------------------------------------------------------------------------- >> >> // setup the OpenSG Glut window >> window = GLUTWindow::create(); >> window->setId(winid); >> window->init(); >> >> // create the cluster server >> server = newClusterServer(window,name,connectionType,address); >> // start the server >> server->start(); > have you tried specifying a multicast port and group address (the > arguments to the ClusterServer c'tor after address)? > Can you run the server(s) under gdb with an active catchpoint for thrown > exceptions? > > (gdb) catch throw > > That would give the point where the socket exception is being thrown. > Did you try running the testClusterServer/testClusterClient programs in > Multicast mode? > > Cheers, > Carsten > > ------------------------------------------------------------------------------ > uberSVN's rich system and user administration capabilities and model > configuration take the hassle out of deploying and managing Subversion and > the tools developers use with it. Learn more about uberSVN and get a free > download at: http://p.sf.net/sfu/wandisco-dev2dev > _______________________________________________ > Opensg-users mailing list > Opensg-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/opensg-users > ------------------------------------------------------------------------------ uberSVN's rich system and user administration capabilities and model configuration take the hassle out of deploying and managing Subversion and the tools developers use with it. Learn more about uberSVN and get a free download at: http://p.sf.net/sfu/wandisco-dev2dev _______________________________________________ Opensg-users mailing list Opensg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/opensg-users
