You've clearly done the first thing I was going to recommend, which was to run Pharo from a terminal to capture any diagnostic output.

The next step would be strace, which might reveal the vm's thoughts.  Use either the -o option or redirect standard error to a file; strace writes nothing to stdout - you've been warned :)   Once you get the output, it will be verbose, but grep is your friend in that case.  You probably know this, something like

   cat strace-log.txt | grep uuid

could be revealing.  It might show where Cog thinks the library lives.  It's ugly, but if you know where Cog insists on looking, then you can put symlink there and *maybe* get going.  I just hit the problem I am describing (and suspect could be ailing you) on Ubuntu and hacked around it that way. Thanks to Nick for the strace pointer.

If you are not sure if/where the UUID library lives, you can try the locate command, or otherwise scrounge around.  If the library is not on the machine, that might be the problem.

HTH.

Bill