Yes i agree.
I looked at the process on linux reading 4000Mb of memory allocated before the crash happen.
Also before the crash we got all aplyers invalidate, so i think there is an error also in the count of the entering / exiting players.
Leaking memory also in that moment.
