1. spark env
spark-env.sh
1) hadoop path
export HADOOP_HOME=/home/tdtc/app/hadoop/2.7.7
export HADOOP_CONF_DIR=$HADOOP_HOME/etc/hadoop
2) spark worker
使用的内存和cpu core数
export SPARK_WORKER_MEMORY=128m
export SPARK_WORKER_CORES=1
3) recovery(zookeeper)
export SPARK_DAEMON_JAVA_OPTS="-Dspark.deploy.recoveryMode=ZOOKEEPER -Dspark.deploy.zookeeper.url=tdtc201:2181,tdtc202:2181,tdtc203:2181,tdtc204:2181 -Dspark.deploy.zookeeper.dir=/spark2"
recoveryMode
spark.deploy.recoveryMode=ZOOKEEPER
url
spark.deploy.zookeeper.url=tdtc201:2181,tdtc202:2181,tdtc203:2181,tdtc204:2181
directory
储存在zk中的位置
spark.deploy.zookeeper.dir=/spark2
2. slaves
注意：有几个YARN，就设置几个slaves！！！
../conf/slaves
tdtc203
tdtc204