Closed mooseframework closed 10 years ago
I found that if I mix two different neutronics material classes (specified on different blocks) that GDB and totalview both crash (valgrind is able to run an picks up the memory leak)
We've done a bit of testing on this over here and haven't been able to trigger it yet (with some simplified examples). We're still working on it.
I'm not able to replicate the bug with the submitted files. My config is: Mac OS X 10.7.3 PETSc 3.2-p5 MOOSE: 8c570ef9f0e20eb38cccbfc857986a17acdf4545 METHOD=dbg
We have a test for this in MOOSE, and it is valgrind clean. Please see test/material_test/adv_mat_couple_test.i for a very simple example.
Closing the ticket!
I ran my nonlinear_keff_test .i in valgrind on my linux machine. The resulting report found some memory leaks associated with the the declaration of material properties of the same name in different material classes. I asked Cody to look at the valgrind memory heap. Cody asked me to verify that they materials where declared on different blocks.
I am including the heap_summary from valgrind
This could be an issue in Pronghorn, but I suspect it may be present at a higher level.