If a drive fail, and the machine is rebooted, gvinum will not detect the failed drive, and just thinks the plex has the right amount of disks.
Gvinum does not have the same header size on different architectures.
During taste, there might be multiple configurations read on disks, but gvinum might not use the newest.
State of gvinum objects are not always set correctly when events happen, e.g. subdisks becoming stale when nothing is really wrong with the data.
It's possible to break your configuration and panicing the kernel if you're not doing things the “right” way.