We didn't update for almost 2 months because of the ongoing issues with PBCs and so many of our tests were failing that we didn't really watch our libMesh builds for other types of errors.  I'm about ready to start my search and will let you know...

Cody

Here are representative errors in optimized mode (This happens in NemesisIO::read -> ParallelMesh::delete_remote_elements

n_elem: 1105p_n_elem: 4422
n_elem: 1105max_elem_id: 1105
terminate called after throwing an instance of 'std::length_error'
  what():  vector::reserve
terminate called after throwing an instance of 'std::length_error'
  what():  vector::reserve

=====================================================================================
=   BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
=   EXIT CODE: 6
=   CLEANING UP REMAINING PROCESSES
=   YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
=====================================================================================
APPLICATION TERMINATED WITH THE EXIT STRING: Abort trap (signal 6)

In Debug mode we hit the assertion:

*** Warning, This code is untested, experimental, or likely to see future API changes: src/mesh/nemesis_io_helper.C, line 62, compiled Sep 14 2012 at 10:32:56 ***
n_elem: 1105p_n_elem: 4422
Assertion `this->n_elem() == this->parallel_n_elem()' failed.
Assertion `this->n_elem() == this->parallel_n_elem()' failed.
[1] src/mesh/parallel_mesh.C, line 1108, compiled Sep 14 2012 at 10:32:57
Assertion `this->n_elem() == this->parallel_n_elem()' failed.
[2] src/mesh/parallel_mesh.C, line 1108, compiled Sep 14 2012 at 10:32:57
Assertion `this->n_elem() == this->parallel_n_elem()' failed.
[3] src/mesh/parallel_mesh.C, line 1108, compiled Sep 14 2012 at 10:32:57
terminate called after throwing an instance of 'libMesh::LogicError'
  what():  Error in libMesh internal logic
terminate called after throwing an instance of 'libMesh::LogicError'
  what():  Error in libMesh internal logic
terminate called after throwing an instance of 'libMesh::LogicError'
  what():  Error in libMesh internal logic
Stack frames: 13
0: 0   libmesh.dylib                       0x00000001010320df libMesh::print_trace(std::ostream&) + 47
1: 1   libmesh.dylib                       0x0000000101518545 libMesh::ParallelMesh::delete_remote_elements() + 501
2: 2   libmesh.dylib                       0x00000001014f0eee libMesh::Nemesis_IO::read(std::string const&) + 29790
3: 3   libmoose-dbg.dylib                  0x00000001009c17a5 ReadMeshAction::readMesh(std::string const&) + 1765
4: 4   libmoose-dbg.dylib                  0x00000001009c31ba ReadMeshAction::act() + 2618
5: 5   libmoose-dbg.dylib                  0x0000000100989919 ActionWarehouse::executeActionsWithAction(std::string const&) + 57
6: 6   libmoose-dbg.dylib                  0x0000000100989a36 ActionWarehouse::executeAllActions() + 102
7: 7   libmoose-dbg.dylib                  0x0000000100abd904 MooseApp::runInputFile() + 164
8: 8   libmoose-dbg.dylib                  0x0000000100abe5b0 MooseApp::parseCommandLine() + 1152
9: 9   libmoose-dbg.dylib                  0x0000000100abd478 MooseApp::run() + 88
10: 10  moose_test-dbg                      0x000000010008320f main + 79
11: 11  moose_test-dbg                      0x00000001000068c8 start + 52
12: 12  ???                                 0x0000000000000003 0x0 + 3
[0] src/mesh/parallel_mesh.C, line 1108, compiled Sep 14 2012 at 10:32:57
terminate called after throwing an instance of 'libMesh::LogicError'




On Fri, Sep 14, 2012 at 11:13 AM, Paul T. Bauman <ptbauman@gmail.com> wrote:


On Fri, Sep 14, 2012 at 12:10 PM, Cody Permann <codypermann@gmail.com> wrote:
Do you have any regression tests for nemesis format?  I haven't been able to run our nemesis tests since our big libMesh update two days ago.  I tried with both the nemesis reader and with mesh.read() without success.  I'm failing assertions in debug mode and seeing either odd library errors in opt mode or having it crash my machine after chewing up many gigabytes of memory during the read.

How long since you updated? What are the errors? A couple of months ago I worked on the output to support vector-valued variables, but we have no nemesis tests, so it's possible I broke something and didn't know it. FWIW, I used Exodus as my primary testbed.