Hi, I'm having almost the exact same problem (this rowcounter jar is one I
compiled myself)
./bin/hadoop jar rowcounter.jar org.myorg.RowCounter /user/myUser/output/
TABLE_NAME
09/06/10 19:44:19 INFO mapred.TableInputFormatBase: split:
0->domain:,18226263
09/06/10 19:44:19 INFO mapred.TableInputFormatBase: split:
1->domain:18226263,
09/06/10 19:44:19 INFO mapred.JobClient: Running job: job_200906101852_0004
09/06/10 19:44:20 INFO mapred.JobClient: map 0% reduce 0%
09/06/10 19:44:24 INFO mapred.JobClient: Task Id :
attempt_200906101852_0004_m_000000_0, Status : FAILED
java.io.IOException: Split class org.apache.hadoop.hbase.mapred.TableSplit
not found
at org.apache.hadoop.mapred.MapTask.run(MapTask.java:314)
After reading through this forum I:
-fiddled with my HADOOP_CLASSPATH in hadoop/conf/hadoop-env.sh
-tried restarting hbase and hadoop, and finally,
-tried moving hbase-0.19.2.jar to the hadoop/lib folder of all the slave
machines.
Still not working though, what exactly was required to get this to work?
Cheers,
Mike
llpind wrote:
>
> Hey Andy,
>
> Yeah I eventually did that instead. It just bothers me why it doesn't
> work otherwise. Anyway, thanks to all
>
>
> Andrew Purtell-2 wrote:
>>
>> For what it's worth, I deploy HBase jars into the Hadoop library
>> directory so I don't have to deal with this. Way back I had classpath
>> problems with 'hadoop jar', and not just with HBase. That may not be a
>> satisfactory answer, but it works for me.
>>
>> - Andy
>>
>>
>
>
--
View this message in context: http://www.nabble.com/PerformanceEvaluation-test-tp23946437p23973184.html
Sent from the HBase User mailing list archive at Nabble.com.
|