Interesting; Did you dive into the MR job history on YARN to see what's the difference like mapper/reducer number and logs? The data size is fetched from MR job counter. Try to compare these jobs to see what's special there.
Best regards, Shaofeng Shi 史少锋 Apache Kylin PMC Email: shaofeng...@apache.org Apache Kylin FAQ: https://kylin.apache.org/docs/gettingstarted/faq.html Join Kylin user mail group: user-subscr...@kylin.apache.org Join Kylin dev mail group: dev-subscr...@kylin.apache.org wonder87 <swordrain2...@hotmail.com> 于2019年11月19日周二 上午2:04写道: > I have a cube A which generate 828kb in the step 4 (extract fact table > distinct columns) > another cube B is the cube cloned by cube A,however ,this cube generate > 2.6G in the step 4.Both of them are same strure and have same previous step > (step1,2,3) > > <http://apache-kylin.74782.x6.nabble.com/file/t930/61574069622_.jpg> > > <http://apache-kylin.74782.x6.nabble.com/file/t930/71574069622_.jpg> > > -- > Sent from: http://apache-kylin.74782.x6.nabble.com/ >