Package org.apache.flink.autoscaler.tuning

Class MemoryTuning

java.lang.Object

org.apache.flink.autoscaler.tuning.MemoryTuning

public class MemoryTuning
extends java.lang.Object

Tunes the TaskManager memory.

Field Summary

Fields

Modifier and Type	Field	Description
static org.apache.flink.runtime.util.config.memory.ProcessMemoryUtils<org.apache.flink.runtime.util.config.memory.taskmanager.TaskExecutorFlinkMemory>	FLINK_MEMORY_UTILS

Constructor Summary

Constructors

Constructor	Description
MemoryTuning()

Method Summary

Modifier and Type	Method	Description
static org.apache.flink.configuration.MemorySize	getTotalMemory(org.apache.flink.configuration.Configuration config, JobAutoScalerContext<?> ctx)
static ConfigChanges	tuneTaskManagerMemory(JobAutoScalerContext<?> context, EvaluatedMetrics evaluatedMetrics, JobTopology jobTopology, java.util.Map<org.apache.flink.runtime.jobgraph.JobVertexID,ScalingSummary> scalingSummaries, AutoScalerEventHandler eventHandler)	Emits a Configuration which contains overrides for the current configuration.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

FLINK_MEMORY_UTILS

public static final org.apache.flink.runtime.util.config.memory.ProcessMemoryUtils<org.apache.flink.runtime.util.config.memory.taskmanager.TaskExecutorFlinkMemory> FLINK_MEMORY_UTILS

Constructor Detail

MemoryTuning

public MemoryTuning()

Method Detail

tuneTaskManagerMemory

public static ConfigChanges tuneTaskManagerMemory(JobAutoScalerContext<?> context,
                                                  EvaluatedMetrics evaluatedMetrics,
                                                  JobTopology jobTopology,
                                                  java.util.Map<org.apache.flink.runtime.jobgraph.JobVertexID,ScalingSummary> scalingSummaries,
                                                  AutoScalerEventHandler eventHandler)

Emits a Configuration which contains overrides for the current configuration. We are not modifying the config directly, but we are emitting ConfigChanges which contain any overrides or removals. This config is persisted separately and applied by the autoscaler. That way we can clear any applied overrides if auto-tuning is disabled.

getTotalMemory

public static org.apache.flink.configuration.MemorySize getTotalMemory(org.apache.flink.configuration.Configuration config,
                                                                       JobAutoScalerContext<?> ctx)