Skip to content

feat: route InfiniBand lifecycle operations by fabric #3531

Description

@poroh

Is this a new feature, an enhancement, or a change to existing functionality?

Change

How would you describe the priority of this feature request

Medium

Please provide a clear description of problem this feature solves

The IB monitor contains partial multi-fabric behavior, but startup rejects multiple fabrics and several runtime paths still construct a client for default.

Removing the startup restriction alone could send mutations to the wrong UFM.

Feature Description

NICo shall route every partition and GUID operation to its owning or observed fabric.

Describe your ideal solution

Remove the single-fabric startup restriction and update:

  • partition reconciliation;
  • QoS updates;
  • partition deletion;
  • monitor partition translation;
  • instance release;
  • force deletion;
  • administrative cleanup.

Key pkey lookups by (fabric_id, pkey). Validate that a requested partition belongs to the fabric where the selected machine GUID was observed.

Treat a GUID reported by multiple fabrics as an error.

Describe any alternatives you have considered

No response

Additional context

No response

Code of Conduct

  • I agree to follow NVIDIA Infra Controller's Code of Conduct
  • I have searched the open feature requests and have found no duplicates for this feature request

Metadata

Metadata

Assignees

No one assigned

    Labels

    Fields

    No fields configured for Enhancement.

    Projects

    Status
    Triage

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions