Skip to content

SDN20-01: IMEX rejoins its domain automatically after a node reboot (host model) #601

Description

@jkenyon-nvidia

Test statement

IMEX (host model): After an unassisted reboot of a node whose service manager owns the IMEX daemon, the IMEX service returns to service automatically and rejoins its domain with no tenant or operator intervention.

Source: docs/test-plan.yaml, SDN20-01.

What to assert

  • Reboot affirmatively confirmed by comparing uptime across the reboot. Never infer that a reboot happened from reachability alone: a node that never went down is reachable too.
  • After boot, the service returns to service and the node rejoins its domain with no tenant or operator intervention.
  • Bounded but generous timeout, since bare-metal GB200-class reboots take minutes.
  • Report elapsed time from boot to domain membership on pass and on fail, so a platform that technically recovers but takes far too long is visible.

Step output contract

The check consumes normalized JSON from a provider step. All parsing of vendor
output happens in the provider script; the validation sees only this shape.

{
  "success": true,
  "platform": "provider",
  "node_id": "node-01",
  "persistence_configured": true,
  "reboot_confirmed": true,
  "uptime_seconds": 94,
  "post_reboot": {
    "service_ready": true,
    "domain_member": true,
    "elapsed_seconds": 61,
    "intervention_required": false
  }
}

Scope

In scope: rebooting one member of the domain and confirming the IMEX service returns and rejoins with no intervention.

Out of scope: whole-domain re-formation from cold. Rebooting the entire domain is not an acceptable substitute for rebooting a single member.

This test is destructive and slow. It reboots a node and must run only on an idle or drained domain. If it carries a slow label, that label must not exclude it from a certification run by default.

Notes

This test proves boot persistence by actually rebooting, rather than reading back that persistence is configured. Note the strict pass condition: a node shipped with IMEX not configured to start at boot fails here, which is deliberate, since shipping it enabled is the provider obligation worth checking.

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    Status
    Todo

    Milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions