How do I provide a full stack trace again? BTW: I did some testing... and even converting the mesh to xda then running the simulation using the xda mesh causes it to hit the same problem... so it's not just Exodus.
ParallelMesh still isn't on by default in SVN right? Derek On Mon, Jun 2, 2008 at 3:27 PM, Roy Stogner <[EMAIL PROTECTED]> wrote: > > On Mon, 2 Jun 2008, Derek Gaston wrote: > >> [0] /Users/gastdr/projects/libmesh/include/geom/remote_elem.h, line >> 91, compiled Jun 2 2008 at 15:01:29 >> terminate called after throwing an instance of 'libMesh::LogicError' >> what(): Error in libMesh internal logic >> p0_19037: p4_error: interrupt SIGx: 6 > >> This is running a "devel" compiled binary in serial. > > This is incredibly odd; nothing should ever be linking to a > remote_elem object unless you're using ParallelMesh in parallel - and > even then, nothing should be calling one of its member functions. > Could you give the full stack trace? > >> Assertion `global_index_map.count(neighbor)' failed. > > This could be the same problem, if neighbor == remote_elem. Let's try > to fix it in serial first and see if the fix works in parallel. > >> If it helps at all I'm reading an Exodus mesh from file. I've looked >> at both of the lines... and can't see what's going wrong... it almost >> looks as if the mesh is corrupt somehow. > > Maybe, but even if it is, there's a bug somewhere. While there are > lots of places where a corrupt mesh might fail a libmesh_assert(), > that remote_elem.h code isn't one of them. > --- > Roy > ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Libmesh-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libmesh-users
