I enabled Yarn Log aggregation using Spark on Cloudera, but when the spark job failed, I run
yarn logs -applicaitonId <id>
or
yarn logs -applicationId <id> -am ALL
All I got are working logs in stderr
But in Yarn failure diagnose line, I managed to see
Diagnostics: User class threw exception: java.lang.reflect.InvocationTargetException
The tricky part is I can't find this exception anywhere, including the /var/log/hadoop-yarn log folder location on each machine as well.
I need this exception details so that I could understand which Jar is missing or what exactly is going wrong with the spark environment settings.
Any ideas how I can find this log location? I found some similar questions with below logs:
17/08/16 02:28:06 ERROR yarn.ApplicationMaster: User class threw exception: java.lang.reflect.InvocationTargetException
java.lang.reflect.InvocationTargetException
at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:62)
at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
at java.lang.reflect.Constructor.newInstance(Constructor.java:423)
Hence I was trying to locate the ApplicationMaster log location and maybe that's different from what we usually do on container logs.