spark-taskscheduler-schedulingmode.adoc

Scheduling Mode — spark.scheduler.mode

Scheduling Mode (aka order task policy or scheduling policy or scheduling order) defines a policy to sort tasks in order for execution.

The scheduling mode schedulingMode attribute is a part of the TaskScheduler Contract.

The only implementation of the TaskScheduler contract in Spark — TaskSchedulerImpl — uses spark.scheduler.mode setting to configure schedulingMode that is merely used to set up the rootPool attribute (with FIFO being the default). It happens when TaskSchedulerImpl is initialized.

There are three acceptable scheduling modes:

• FIFO with no pools but a single top-level unnamed pool with elements being TaskSetManager objects; lower priority gets Schedulable sooner or earlier stage wins.

• FAIR with a hierarchy of Schedulable (sub)pools with the rootPool at the top.

• NONE (not used)

Note	Out of three possible SchedulingMode policies only FIFO and FAIR modes are supported by TaskSchedulerImpl.

Note	After the root pool is initialized, the scheduling mode is no longer relevant (since the Schedulable that represents the root pool is fully set up). The root pool is later used when TaskSchedulerImpl submits tasks (as TaskSets) for execution.

Note	The root pool is a Schedulable. Refer to Schedulable.

Monitoring FAIR Scheduling Mode using Spark UI

Caution

FIXME Describe me…​