Weird seeing nothing related in the drillbit.log, the only extra thing I can see is "SYSTEM ERROR: Drill Remote Exception" in the profile after the query fails.
The only other thing I can think of at this stage is the local Disk on this new instance is at 91% although I am writing to HDFS so would not expect that to be an issue. Will have to keep digging. Thanks On Wed, Aug 17, 2016 at 9:01 PM Jinfeng Ni <[email protected]> wrote: > The source code that raised this exception seems to be here [1]. Can > you please check your drillbit.log and see if it has more information? > > > [1] > https://github.com/apache/drill/blob/master/exec/java-exec/src/main/java/org/apache/drill/exec/physical/impl/BaseRootExec.java#L144 > > On Wed, Aug 17, 2016 at 8:37 PM, Stefan Sedich <[email protected]> > wrote: > > I have a gzipped json sample file ~2GB which I can create a parquet table > > from perfectly fine on my laptop. > > > > I have spun up a new EMR cluster running MapR M5 and am using the > bootstrap > > script: https://www.mapr.com/blog/bootstrap-apache-drill-amazon-emr > > > > Running the same CTAS I can see if gets through reading around 1/5th of > the > > rows from the json and crashes with the following: > > > > Error: SYSTEM ERROR: RuntimeException: Error closing operators > > > > What is my best way to really diagnose what is going on? I wouldn't have > > thought it would be a memory issue as these EMR nodes have far more > memory > > that that of my laptop. > > > > Any advice is appreciated. > > > > > > > > Thanks >
