Hive-Mapreduce-Join

[fasi meg <fa...@yahoo.com>]

Hello All,
 
I recently started working with Hive and Mapreduce and I was trying to join my star schema tables(1 fact table with 9 dimensions).
The problem I have is well…. Mapreduce works fine when I ran Select count(*) from table;
 
But MR couldn't possibly finish the join task...I attached the syntax I used to join the tables(I edit it so that you can see the steps clearly).
I also tried only to join two tables, but still it doesn't have any error from the console …..it keeps showing me the following message until I stop it.
2013-08-29 11:47:10,476 Stage-1 map = 52%,  reduce = 15%, Cumulative CPU 1213.76 sec
 
Again I changed my method and used:
CREATE TABLE Join_1 AS
   SELECT
          t1.col1
         ,t1.col2
         ,t1.col3
         ,t1.col5
         ,.
        ,t2.col2
FROM facttable t1 LEFT OUTER JOIN dim t2 ON (t1.col1 = t2.col1); but still the problem is the same
 
I tried to fix Circular redirect error based on the solution provided here http://svn.apache.org/viewvc/hadoop/common/trunk/hadoop-mapreduce-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-web-proxy/src/main/java/org/apache/hadoop/yarn/server/webproxy/WebAppProxyServlet.java?r1=1295314&r2=1295313&pathrev=1295314
But it didn’t fix the error.
My question:
1.       Any idea how to fix CircularRedirect error?
2.       Every time I ran different joins  It always uses 27 number of mappers and 7 number of reducers. How do I change that? I set mapreduce.tasktracker.map/reduce.tasks.maximum to 2 but no change.
3.       One more question how can I make sure that MR is using full capacity of the resources I have?
 
I attached all the log files and important files I could possibly think of and I hope I will receive a response from you soon.
Hadoop Version  0.23.1
Hive Version 0.10.0
I am using MR2(YARN)
 
Thank you in advance,
With best regards,
Fasika