解决Spark On Yarn执行中executor内存限制问题
集群版本 Spark1.3.1 + Hadoop.2.6-CDH5.4
启动Spark-SQL的JDBC服务,如下:
start-thriftserver.sh --master yarn-client --executor-memory 15g --num-executors 13
提示内存不足
Required executor memory (xxx MB) is above the max threshold (xxx MB) of this cluster!