Skip to content

Improve resource manager observability #3197

@sukunrt

Description

@sukunrt

We should provide more inspection tools for users to make it easier for users to inspect resource manager usage. I have 3 specific items in mind.

  • Provide a way to print resource usage: Add a way to print resource limits and current consumption #3193
  • Add prometheus metrics for the configured limits so these can be added to dashboards for convenient checking of the upper bound.
  • Log the configured rcmgr limits on startup
  • Improve the documentation for Limits
    • What's a PartialLimitConfig
    • What's a ScalingLimitConfig
    • What's a default LimitVal which depends on context
    • Move a lot of the README stuff to the actual objects so godoc can pick it up
    • Add many more examples for common things people want
  • Simplify the API for adjusting the default resources allocated. By default, an 1/8 of the resources are allocated. It should be easy enough to adjust this multiplier.

cc @2color

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions