问题:因为之前hadoop集群上配了lzo,spark还没有配置
方法:在spark的spark-env.sh中配置lzo(将lzo的jar包路径给spark)
export SPARK_LIBRARY_PATH=$SPARK_LIBRARY_PATH:/opt/module/hadoop-2.7.2/lib/native
export SPARK_CLASSPATH=$SPARK_CLASSPATH:/opt/module/hadoop-2.7.2/share/hadoop/common/hadoop-lzo-0.4.20.jar