Since our data warehouse data source is Kafka, the offline data warehouse needs to use Flume to collect data from Kafka to HDFS.

In actual projects, it is not feasible to always start Flume tasks in Xshell. Firstly, because there are many Flume tasks in the project, and secondly, once the Xshell page is closed, the Flume tasks will stop, which is very inconvenient. Therefore, it is necessary to start Flume tasks in the background.