Spark 入门之四：Spark任务调度架构

1. 以Spark集群的方式运行
   
   
   1. standlone 的运行：bin/spark-shell --master spark://hadoop1:7077 --executor-memory 3g
2. 运行在Yarn上
   
   
   1. yarn-client的运行：bin/spark-shell --master yarn-client --executor-memory 3g --num-executors 3
   2. yarn-cluster的运行：./bin/spark-submit --master yarn-cluster --class week2.SogouQA --executor-memory 3g week2.jar hdfs://hadoop1:8000/dataguru/data/SogouQ1.txt  hdfs://hadoop1:8000/dataguru/week2/output3