The shell script generates a Makefile and the Makefile runs the hadoop commands, so that the parallel dep handling is entirely handed off to Make.
This make it super easy to run 2 parallel workloads at all times - unlike xargs -P 2, this is much more friendly towards complex before/after deps and failure handling.
https://github.com/hortonworks/hive-testbench/blob/hive14/tp...
The shell script generates a Makefile and the Makefile runs the hadoop commands, so that the parallel dep handling is entirely handed off to Make.
This make it super easy to run 2 parallel workloads at all times - unlike xargs -P 2, this is much more friendly towards complex before/after deps and failure handling.