You can look at the query plan in the Web UI and take note of memory consumption for different Fragments and Operators to get an idea of which one is consuming memory on the node(s).
You can look at the Performance Tuning documentation for some guidance to tune and set parameters to help with your issue. https://drill.apache.org/docs/performance-tuning-introduction/ <https://drill.apache.org/docs/performance-tuning-introduction/> —Andries > On Oct 27, 2015, at 1:01 PM, Rajesh Balel -X (rbalel - COGNIZANT TECHNOLOGY > SOLUTIONS US CORPORATION at Cisco) <[email protected]> wrote: > > Hello All, > > We are experiencing below error message in particular while trying to run > "select" on 1 billion record hive table join with 5 million > > Error: RESOURCE ERROR: One or more nodes ran out of memory while executing > the query. > Fragment 1:169 > [Error Id: 23a044a5-e94e-4467-81b2-ab2cb23a5be7 on hdnprd-c01-r01-13:31010] > (state=,code=0) > > Is there some recommended memory settings / configuration parameters that can > be implemented to pass through this error? > > Please confirm > > Thanks > Rajesh Balel > [email protected]<mailto:[email protected]> > Enterprise Hadoop Application Support >
