I want to operate hadoop on the shared filesystem such as NFS, Lustre. I think locality calculation in this scenario is meaningless because local disk is not used.
Is there any configuration parameters not to use locality calculation? or I would like to know how to adjust locality calculation in hadoop which is optimized for the shared filesystem? I am suspecting locality wait factor (mapreduce.job.locality.wait.factor) is regarding to this, but I am not sure.
Any suggestions?
