Hadoop调度程序与Oozie

Question

What is the difference between oozie and Hadoop schedulers (FIFO, Fair, Capacity, Delay etc) ? Which one is normally used in production environments ?

Answer 1

They offer different functions.

Oozie is a job scheduler and orchestration engine. It'll execute YARN jobs within the configured queue. For example, mapreduce.job.queuename in MR/Java action, tez.queue.name in Hive Action (on Tez), or --queue in Spark.

The queue you use depends on the workloads you have, but Fair or Capacity work best for multi-tenant workloads.

FIFO ordering within a queue works best if you have jobs that need to run before others, although Oozie has its own definition of how to order a full job

Also: Job and Task Scheduling In Hadoop