Adding -Dmapred.job.tracker=local to your command line will cause a
map-reduce program to be executed entirely locally. This may not be what
you mean.
You can run a map-reduce program on a machine outside of the hadoop cluster
by just copying the configuration files to the machine where you want to run
your program. Your program will contact the namenode and job tracker to
access or store data and to start tasks. You can get away with very little
of the normal distribution, but I find it easiest to copy the entire
distribution to the machine that runs the program.
On 1/22/08 8:01 PM, "ma qiang" <
maqiang1984@...> wrote:
> Dear colleagues:
> I have some mapreduce codes in hadoop,now I want to invoke
> these codes in another place outside of hadoop server. Who can gell me
> how?
> Thanks !
