[YUNIKORN-847] Use resource weighting for node sorting

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.12.1
• Component/s: core - scheduler
• Labels: pull-request-available
• Target Version: 0.12.1
• Assignee: Craig Condit
• Reporter: Craig Condit
• Created: 15/Sep/21 19:19
• Updated: 21/Jan/22 21:36
• Resolved: 17/Sep/21 22:19

Description

After YUNIKORN-807, node sorting uses a scoring algorithm which uses the largest resource share to determine priority. However, resource units are not taken into account, and consequently, either fair or binpacking tends to look almost exclusively at the largest absolute values for units. In some environments this can lead to priority inversions.

We should instead calculate in-use percentages for each resource, and then weight them to arrive at an overall usage percentage, which can then be used for scoring. Weights can be controlled at the partition level.

 Configuration is as follows:

partitions:
  - name: default
    nodesortpolicy:
      type: fair
      resourceweights:
        vcore: 4.0
        memory: 1.0
    queues:
      - name: root
        submitacl: '*'

 In this example, vcore is weighted double what memory is. If resourceweights is not specified, the configuration is equivalent to:

{ vcore: 1, memory: 1}

 

For example, in the default configuration, if a node has 60% of its vcores in use, and 20% of its memory in use, node usage would be calculated at 40% using equal weights. In the example given above, usage would be calculated as 52%.

 

 

Issue Links

• is required by: YUNIKORN-852 Document changes from YUNIKORN-847
• relates to: YUNIKORN-807 Improve performance of node sorting
• supercedes: YUNIKORN-831 node sorting should check other resources of nodes instead of comparing node id directly
• supercedes: YUNIKORN-21 Optimize node sorting algorithms
• links to: GitHub Pull Request #323

Sub-Tasks

• Add "resource weighting" to /ws/v1/partitions: Closed
• update the docs in web site about "nodeSortingPolicy": Closed