protected static class GenericResourceAwareSchedulingPriorityStrategy.GrasSimulatedUser extends DefaultSchedulingPriorityStrategy.SimulatedUser
Modifier and Type | Field and Description |
---|---|
Map<String,Double> |
guaranteedGenericResources |
assignedCpu, assignedMemory, guaranteedCpu, guaranteedMemory, tds
Constructor and Description |
---|
GrasSimulatedUser(User other,
ISchedulingState cluster) |
Modifier and Type | Method and Description |
---|---|
protected double |
getScore(double availableCpu,
double availableMemory,
Map<String,Double> availableGenericResources) |
protected double |
getScore(double availableCpu,
double availableMemory,
Map<String,Double> availableGenericResources,
TopologyDetails td)
Get a score for the simulated user.
|
TopologyDetails |
simScheduleNextHighest() |
getNextHighest, getScore, getScore
public GrasSimulatedUser(User other, ISchedulingState cluster)
public TopologyDetails simScheduleNextHighest()
simScheduleNextHighest
in class DefaultSchedulingPriorityStrategy.SimulatedUser
protected double getScore(double availableCpu, double availableMemory, Map<String,Double> availableGenericResources, TopologyDetails td)
Get a score for the simulated user. This is used to sort the users, by their highest priority topology. Only give user guarantees that will not exceed cluster capacity. Score of each resource type is calculated as: (Requested + Assigned - Guaranteed)/clusterAvailable The final score is a max over all resource types. Topology score will fall into the following intervals if: User is under quota (guarantee): [(-guarantee)/available : 0] User is over quota: (0, infinity) Unfortunately, score below 0 does not guarantee that the topology will be scheduled due to resources fragmentation.
availableCpu
- available CPU on the cluster.availableMemory
- available memory on the cluster.availableGenericResources
- available generic resources (other that cpu and memory) in clustertd
- the topology we are looking at.Copyright © 2020 The Apache Software Foundation. All rights reserved.