I want to run Mahout's K-Means example in a hadoop cluster of 5 machines. Which Mahout jar files should I need to keep in all the nodes, in order for the K-Means to be executed in a distributed manner. Thanks.
-Venkiram
Asked
Active
Viewed 1,338 times
1 Answers
2
If you really just want to run the built-in K-Means, or other jobs with static drivers, the answer is 'none'. The mahout 'job' jars are self-contained hadoop job jars. If you submit a job to the cluster with 'hadoop job' it will work without any other jars.
bmargulies
- 97,814
- 39
- 186
- 310
-
Thanks. I have installed Mahout jars in master node, should I replicate it in all slave nodes also? – Venkiram Apr 15 '11 at 14:47
-
Now you are beyond what I know about hadoop. – bmargulies Apr 16 '11 at 11:27
-
1Isn't the job driver supposed to take care of this by localizing the .jar on all the nodes it uses? – Has QUIT--Anony-Mousse Sep 26 '12 at 18:23