Can't get cube segment size

[xingfengshen]

I build test_kylin_cube_with_slr_empty failed, it shows that can't get cube segment size. The error log as followed:
#13 Step Name: Load HFile to HBase Table

Start to execute command:?
?-input /tmp/kylin-e3f73338-edb8-4a1c-9013-e14d61f58f0c/test_kylin_cube_with_slr_empty/hfile/ -htablename KYLIN_QA_CUBE_HYIZYTA36N -cubename test_kylin_cube_with_slr_empty
Command execute return code 0

Failed with Exception:java.lang.RuntimeException: Can't get cube segment size.
at com.kylinolap.job.flow.JobFlowListener.updateCubeSegmentInfoOnSucceed(JobFlowListener.java:245)
at com.kylinolap.job.flow.JobFlowListener.jobWasExecuted(JobFlowListener.java:99)
at org.quartz.core.QuartzScheduler.notifyJobListenersWasExecuted(QuartzScheduler.java:1985)
at org.quartz.core.JobRunShell.notifyJobListenersComplete(JobRunShell.java:340)
at org.quartz.core.JobRunShell.run(JobRunShell.java:224)
at org.quartz.simpl.SimpleThreadPool$WorkerThread.run(SimpleThreadPool.java:573)

other steps are successful, which model will set the segment size?

[binmahone]

job engine will retrieve the stats from hadoop log after the cube building is done. it seems the job engine failed to find the log it needed. please check the log output of a job step called "Convert Cuboid Data to HFile" (see attached image) to see if there's anything like

HDFS: Number of bytes read=3851451742573

[binmahone]

btw, if you could leave us your contact info, we might be able to communicate directly with you, so that we could better understand your requirements.

[xingfengshen]

Hi, we have fixed this issue.

1st: we start JobHistoryServer
java.net.ConnectException: to 0.0.0.0:10020 failed on connection exception
config hadoop的/etc/hadoop/mapred-site.xml file

mapreduce.jobhistory.address
host:10020

execute this cmd:
mr-jobhistory-daemon.sh start historyserver

2nd: because HBase-0.98.0-hadoop2 use hadoop 2.2.0, but in our cluster the hadoop version is 2.4.1, so it will cause an exception, details can see https://issues.apache.org/jira/browse/MAPREDUCE-5831, so we replace the hadoop jars to 2.4.1

now we build the cube successful.

Thank you very much!

[lukehan]

Hi @xfhap, this is really great reference for others to setup their cluster.

Thank you very much.
Luke