KYLIN simply use a utility class ToolRunner to submit MR job. As for your question, you may need to check your cluster’s configuration or do a health check to see if YARN can see all the nodes.
Best Regard Zhou QianHao On 5/25/15, 6:51 PM, "190184501" <[email protected]> wrote: >hi , when i run kylin_sales_cube with hadoop cluster ， or other cube, > > >i found kylin'job only run in a single hadoop node, > > >eg: >hadoop1(192.168.1.11) is master >hadoop2(192.168.1.12) is slave >hadoop3(192.168.1.13) is slave >and when kylin_sales_cube is bulid, i see the kylin job is assigned to >yarn and only hadoop2 run the mr's job > > >so, my question is : > > >why only one node --- hadoop2 run the mr's job, not hadoop2 hadoop3 run >kylin's mr job ?? > > >how kylin distribute job to hadoop yarn?? > > >is something i forget to write in my Configuration file ?? if is , >where > > >thx.
