Major Description
Currently, the configure-sqoop scripts that checks the value of various _HOME variables (e.g. HADOOP_MAPRED_HOME, HCAT_HOME, etc.) has the following heuristic:
1. Use the *_HOME variable's value, if defined.
2. If not, assign /usr/lib/<relevant location> to the variable.
3. If /usr/lib/<relevant location> doesn't exist, error out
I would like to propose to modify this to add one more case before we error out.
1. Use the *_HOME variable's value, if defined.
2. If not, assign /usr/lib/<relevant location> to the *_HOME variable.
3. If /usr/lib/<relevant location> doesn't exist, check for hadoop and hcat at the same directory level as sqoop
4. If #3 fails as well, error out.
This is backwards compatible because it wouldn't be touching any of the existing cases. It will just add more location to check for presence of hadoop, hcat etc. before erroring out.