2019-12-12 09:37:43
Spark Shell
Spark-shell is Shell Spark comes with an interactive program, user-friendly interactive programming, you can write Scala program in command Spark
Spark-shell used for Spark-Shell test has two modes start local mode and a mode Cluster
Local mode:
ps: Spark must enter the installation directory under the bin directory spark-shell start is the local mode, local mode only SparkSubmit start a process in the machine, there is no cluster to establish contact, although there SparkSubmit process but will not be submitted to the cluster
Cluster mode:
Spark must enter the installation directory bin directory under the Spark-shell \ --master the Spark: // hadoop01: 7077 \ --executor-Memory 512M \ --total-Executor-Cores 1 PS: there must be a command --master the latter two can not
Exit Shell:
1. Use: quit exit the shell 2. Use ctrl + c to exit the shell PS: use ctrl + c to exit the shell may occupy the port of background appears to view the listening port netstat -apn | grep 4040 as long as the port occupied by the process will kill
Comparison of Spark2.2shell and Spark1.6shell
1) Spark2.2
2) Spark1.6
Description: Spark2.X version shell has two built-in objects
SparkContext ——> 变量名:sc
SparkSession ——>变量名:spark
SparkSession不能单独拿出来解释,2.x之后spark将SQLContext和HiveContext进行整合提供一种全新的创建方式SparkSession
Spark1.6版本shell中有两个内置对象
SparkContext ——>变量名:sc
SQLContext ——>变量名:sqlcontext