Configuring Dynamic Resource Scheduling in Yarn Mode

Scenario

Resources are a key factor that affects Spark execution efficiency. When a long-running service (such as the JDBCServer) is allocated with multiple executors without tasks but resources of other applications are insufficient, resources are wasted and scheduled improperly.

Dynamic resource scheduling can add or remove executors of applications in real time based on the task load. In this way, resources are dynamically scheduled to applications.

Procedure

1. Configure the external shuffle service.
2. Log in to FusionInsight Manager and choose Cluster > Services > Spark. Click Configurations then All Configurations. Enter spark.dynamicAllocation.enabled in the search box and set its value to true to enable the dynamic resource scheduling function. This function is disabled by default.

Table 1 lists some optional configuration items.

Table 1 Parameters for dynamic resource scheduling

Configuration Item	Description	Default Value
spark.dynamicAllocation.minExecutors	Indicates the minimum number of executors.	0
spark.dynamicAllocation.initialExecutors	Indicates the number of initial executors.	0
spark.dynamicAllocation.maxExecutors	Indicates the maximum number of executors.	2048
spark.dynamicAllocation.schedulerBacklogTimeout	Indicates the first timeout period for scheduling.	1s
spark.dynamicAllocation.sustainedSchedulerBacklogTimeout	Indicates the second and later timeout interval for scheduling.	1s
spark.dynamicAllocation.executorIdleTimeout	Indicates the idle timeout interval for common executors.	60s
spark.dynamicAllocation.cachedExecutorIdleTimeout	Indicates the idle timeout interval for executors with cached blocks.	JDBCServer: 2147483647s IndexServer: 2147483647s SparkResource: 120

The external shuffle service must be configured before using the dynamic resource scheduling function.