The idea is to find the bands and to calculate their Berry phase associated or other geometrical or topological objects which could give you invariants such as the Chern numbers. Once you have the topological invariants, this means that the transport of spins or other magnetic configurations, don't depends of local interactions as could be the usual defects of the lattice.
In any case I understand that my answer is too conceptual and far of providing you with tools to follow it. I give you one link for going to a place where you can find more details:
With this information I suppose three different situations:
1) Avogadro is a specific "visualising program" with the "extra tool" of the perform a optimization with a force field. The results obtained with this method are not too much reliable in a study but it is very useful to build a very complex structures (i.e with gaussview is much more complex draw some type of compounds). This tool has a consumption of RAM and CPU and a possibility of error is that the system do not has enough resources if the molecule is too much large.
2)If using the program only to build inputs, then maybe you has a problem reading the outputs or preparing the inputs. If you use this software in this way, please report it with more technical details.
3) Also, some kind of system, such crystals, can give error depending where you read the file or if the crystal structures is not properly introduced. For example, a common "bug" happens when you try to add cells to your crystal. This happens because some crystal file only has a portion of the unit cell that Avogadro interpret.
In general, if Avogadro gives you a problem, you could change to other program such Molden or MacMolPLT (which this last one is particular very useful to prepare inputs to gamess software).