
上QQ阅读APP看书,第一时间看更新
Standalone
Standalone mode uses a basic cluster manager that is supplied with Apache Spark. The spark master URL will be as follows:
Spark://<hostname>:7077
Here, <hostname> is the name of the host on which the Spark master is running. We have specified 7077 as the port, which is the default value, but this is configurable. This simple cluster manager currently supports only FIFO (first-in-first-out) scheduling. You can contrive to allow concurrent application scheduling by setting the resource configuration options for each application; for instance, using spark.core.max to share cores between applications.