fix: allow non-churn empty nodes to be disrupted

[rschalo]

Fixes #N/A

Description
Today, when Karpenter performs validation for empty nodes, if a single node experience pod churn, the entire batch of empty nodes is considered invalid. This change alters validation such that only the nodes that experience churn are excluded from disruption.

How was this change tested?
Unit tests, also:

I applied this change with a sleep during validation (see "validating candidates" log) to a cluster with two nodes, one pod per node, and then scaled the deployment down to zero. While in the sleep, I scaled up the deployment to one pod. The node the pod scheduled to had its consolidatable status condition removed (practical-tharp-919765395) and the command proceeded with the nodeclaim that was still empty eager-feynman-1734482834.

karpenter-5944796d5f-sf9zg controller {"level":"DEBUG","time":"2025-05-08T17:09:37.713Z","logger":"controller","caller":"disruption/validation.go:123","message":"validating candidates","commit":"3087cce-dirty","controller":"disruption","namespace":"","name":"","reconcileID":"7ee690c9-7c65-476c-860b-37a6e699204d"}

karpenter-5944796d5f-sf9zg controller {"level":"DEBUG","time":"2025-05-08T17:09:40.964Z","logger":"controller","caller":"disruption/consolidation.go:66","message":"removing consolidatable status condition","commit":"3087cce-dirty","controller":"nodeclaim.disruption","controllerGroup":"karpenter.sh","controllerKind":"NodeClaim","NodeClaim":{"name":"arm-2cpu-dxf5m"},"namespace":"","name":"arm-2cpu-dxf5m","reconcileID":"4c049d78-633c-4615-9887-5b12c02af3ed","Node":{"name":"practical-tharp-919765395"}}

karpenter-5944796d5f-sf9zg controller {"level":"INFO","time":"2025-05-08T17:09:47.714Z","logger":"controller","caller":"disruption/controller.go:206","message":"disrupting node(s)","commit":"3087cce-dirty","controller":"disruption","namespace":"","name":"","reconcileID":"7ee690c9-7c65-476c-860b-37a6e699204d","command-id":"6c95a5be-9992-47e5-b6e4-4645e55e03db","reason":"empty","decision":"delete","disrupted-node-count":1,"replacement-node-count":0,"pod-count":0,"disrupted-nodes":[{"Node":{"name":"eager-feynman-1734482834"},"NodeClaim":{"name":"arm-2cpu-k72c5"},"capacity-type":"on-demand","instance-type":"c-1x-arm64-linux"}],"replacement-nodes":[]}

karpenter-5944796d5f-sf9zg controller {"level":"DEBUG","time":"2025-05-08T17:09:48.659Z","logger":"controller","caller":"orchestration/queue.go:236","message":"command succeeded","commit":"3087cce-dirty","controller":"disruption.queue","namespace":"","name":"","reconcileID":"58809c5a-348c-472b-ac72-b0f3e75843bb","command-id":"6c95a5be-9992-47e5-b6e4-4645e55e03db","reason":"Empty","decision":"delete","disrupted-node-count":1,"replacement-node-count":0,"disrupted-nodes":[{"Node":{"name":"eager-feynman-1734482834"},"NodeClaim":{"name":"arm-2cpu-k72c5"}}],"replacement-nodes":[]}

karpenter-5944796d5f-sf9zg controller {"level":"INFO","time":"2025-05-08T17:09:48.683Z","logger":"controller","caller":"terminator/terminator.go:89","message":"tainted node","commit":"3087cce-dirty","controller":"node.termination","controllerGroup":"","controllerKind":"Node","Node":{"name":"eager-feynman-1734482834"},"namespace":"","name":"eager-feynman-1734482834","reconcileID":"cdf22cb6-0686-42f7-be34-68fec7cb6418","NodeClaim":{"name":"arm-2cpu-k72c5"},"taint.Key":"karpenter.sh/disrupted","taint.Value":"","taint.Effect":"NoSchedule"}

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[coveralls]

Pull Request Test Coverage Report for Build 15120678788

Details

• 19 of 26 (73.08%) changed or added relevant lines in 2 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.01%) to 81.947%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
pkg/controllers/disruption/validation.go	16	23	69.57%

Totals
Change from base Build 15075049606:	0.01%
Covered Lines:	10150
Relevant Lines:	12386

---

💛 - Coveralls

[rschalo]

I think that the best way to test this properly would require a refactoring of the disruption controller for injecting the methods rather than adding them directly to the methods member of the controller. Alternatively, validation could be passed as a param and a mocked validator that creates a pod and binds it to a node during validation could be added.

[jmdeal]

/lgtm
/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: jmdeal, rschalo

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [jmdeal]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment