These scripts are used to run MySQL dumps and restores in parallel, with one job per table. This brings significant performance increases in most dumps and restores, especially when SSDs are used. By default, jobs will be spawned for each concurrent thread that the system can run. As well, lbzip2 will be used, as with bzip2 any dumps or restores where a small number of tables contain the majority of the data will be blocked by a single bzip2 process on a CPU core. While this might seem a bit extreme, in practice the OS X scheduler (and presumably the Linux schedule) handles the large number of processes and threads just fine. By default, Ubuntu sets the --tollef flag in /etc/parallel/config. This breaks these scripts in amusing and horrible ways. Remove that line from the global configuration if you want to use these scripts. See GNU parallel not working at all for details.