spark优化——依赖包传入HDFS_spark.yarn.jar和spark.yarn.archive的使用

一、参数说明

启动Spark任务时，在没有配置spark.yarn.archive或者spark.yarn.jars时， 会看到不停地上传jar，非常耗时；使用spark.yarn.archive可以大大地减少任务的启动时间，整个处理过程如下。

二、spark.yarn.archive使用

1.在本地创建zip文件

silent@bd01:~/env/spark$ cd jars/
silent@bd01:~/env/spark$ zip spark2.0.0.zip ./*

注：zip包为全量包

2.上传至HDFS并更改权

silent@bd01:~/env/spark$ /usr/ndp/current/hdfs_client/bin/hdfs dfs -mkdir /tmp/spark-archive
silent@bd01:~/env/spark$ /usr/ndp/current/hdfs_client/bin/hdfs dfs -put ./spark2.0.0.zip /tmp/spark-archive
silent@bd01:~/env/spark$ /usr/ndp/current/hdfs_client/bin/hdfs dfs -chmod 775 /tmp/spark-archive/spark2.0.0.zip.zip

3.配置spark-defaut.conf

spark.yarn.archive  hdfs:///tmp/spark-archive/spark2.0.0.zip

三、spark.yarn.jars使用说明

1. 上传依赖jar包

silent@bd01:~/env/spark$ /usr/ndp/current/hdfs_client/bin/hdfs dfs -mkdir hdfs://bd01/user/asiainfo/jars/
silent@bd01:~/env/spark$ /usr/ndp/current/hdfs_client/bin/hdfs dfs -put ./spark2.0.0.zip hdfs://bd01/user/asiainfo/jars/
silent@bd01:~/env/spark$ /usr/ndp/current/hdfs_client/bin/hdfs dfs -chmod 775 hdfs://bd01/user/asiainfo/jars/spark2.0.0.zip.zip

2.配置spark-defaut.conf

spark.yarn.jars=local:/opt/cloudera/parcels/CDH-6.1.0-1.cdh6.1.0.p0.770702/lib/spark/jars/*,local:/opt/cloudera/parcels/CDH-6.1.0-1.cdh6.1.0.p0.770702/lib/spark/hive/*,hdfs://bd01/user/asiainfo/jars/*.jar

注：本地配置local,hdfs标记为hdfs目录即可