The Node Tuning Operator Updated the Kubeletconfig of the cluster. This resulted to the cluster to start a Machine Config Rollout without the user to do any action. In more detail:

The node tuning operator updates the kubelet-config:
~~~
2023-07-18T21:23:27.048335958+03:00 I0718 18:23:27.048303       1 resources.go:159] Update kubelet-config "performance-worker"
~~~

From kubeletconfig CR.
~~~
    manager: cluster-node-tuning-operator
     operation: Update
     time: "2023-07-18T18:23:27Z"
~~~
A new "99-worker-generated-kubelet" MC is getting created with name "99-worker-generated-kubelet-1" from the Machine Config Controller.

A new rendered MC that includes the new kubeletconfig MC is getting created from the controller.
~~~
2023-07-18T21:26:24.722890720+03:00 I0718 18:26:24.722788       1 render_controller.go:510] Generated machineconfig rendered-worker-0d708d32a1e04ec8b02986c40d10828b from 16 configs: 
[{MachineConfig  00-worker  
machineconfiguration.openshift.io/v1  } {MachineConfig  01-worker-container-runtime  

machineconfiguration.openshift.io/v1  } {MachineConfig  01-worker-kubelet  
machineconfiguration.openshift.io/v1  } {MachineConfig  50-nto-worker  
machineconfiguration.openshift.io/v1  } {MachineConfig  50-performance-worker  
machineconfiguration.openshift.io/v1  } {MachineConfig  99-worker-chrony-conf-override  
machineconfiguration.openshift.io/v1  } {MachineConfig  99-worker-generated-containerruntime  
machineconfiguration.openshift.io/v1  } {MachineConfig  99-worker-generated-kubelet  
machineconfiguration.openshift.io/v1  } {MachineConfig  99-worker-generated-kubelet-1  <---------------------------- This one
machineconfiguration.openshift.io/v1  } {MachineConfig  99-worker-generated-registries  
machineconfiguration.openshift.io/v1  } {MachineConfig  99-worker-legacy-kdump-configuration  
machineconfiguration.openshift.io/v1  } {MachineConfig  99-worker-ssh  
machineconfiguration.openshift.io/v1  } {MachineConfig  coredump-mc-worker  
machineconfiguration.openshift.io/v1  } {MachineConfig  softpanic-worker  
machineconfiguration.openshift.io/v1  } {MachineConfig  worker-custom-timezone-configuration  
machineconfiguration.openshift.io/v1  } {MachineConfig  worker-std-r750-1-combo-sctp  
machineconfiguration.openshift.io/v1  }]
2023-07-18T21:26:24.723116641+03:00 I0718 18:26:24.723021       1 event.go:285] Event(v1.ObjectReference{Kind:"MachineConfigPool", Namespace:"", Name:"worker", UID:"e104aa2e-0729-49ae-8022-502cbb414ab1", APIVersion:"machineconfiguration.
openshift.io/v1", ResourceVersion:"87008874", FieldPath:""}): type: 'Normal' reason: 'RenderedConfigGenerated' rendered-worker-0d708d32a1e04ec8b02986c40d10828b successfully generated (release version: 4.12.8, controller version: 731341b8
9e72d53abb349aff98d09e281e471066)
~~~
This is getting applied to the cluster.

The differences between the previous and the new rendered MC are the below:
~~~
[nstamate@fedora ~]$ diff rendered-worker-0d708d32a1e04ec8b02986c40d10828b-kubelet-config.yaml rendered-worker-1102262599b5c40ca2c2aae79f1b35d0-kubelet-config.yaml 
48c48,54
<   "cpuManagerReconcilePeriod": "0s",
---
>   "cpuManagerPolicy": "static",
>   "cpuManagerPolicyOptions": {
>     "full-pcpus-only": "true"
>   },
>   "cpuManagerReconcilePeriod": "5s",
>   "memoryManagerPolicy": "Static",
>   "topologyManagerPolicy": "single-numa-node",
54a61,66
>   "evictionHard": {
>     "imagefs.available": "15%",
>     "memory.available": "100Mi",
>     "nodefs.available": "10%",
>     "nodefs.inodesFree": "5%"
>   },
65,68c77,80
<   "allowedUnsafeSysctls": [
<     "fs.mqueue.*",
<     "net.*"
<   ],
---
>   "kubeReserved": {
>     "memory": "500Mi"
>   },
>   "reservedSystemCPUs": "0-3,52-55",
79c91,99
<   "shutdownGracePeriodCriticalPods": "0s"
---
>   "shutdownGracePeriodCriticalPods": "0s",
>   "reservedMemory": [
>     {
>       "numaNode": 0,
>       "limits": {
>         "memory": "1100Mi"
>       }
>     }
>   ]
~~~
Then the changes are reverted automatically.
~~~
2023-07-19T10:39:09.804948915+03:00 I0719 07:39:09.804900       1 resources.go:159] Update kubelet-config "performance-worker"
~~~
And the Machine Config Controller targets the previous rendered MC.

A new rollout starts.

The admin didn't touch anything.

Must-gather and other manifests are in the attached case.

N/A

Cannot reproduce as it happened only once.

1.
2.
3.

The Kubelet config is changed by the Node Tuning Operator at random times

The Kubelet config should not be changed by the Node Tuning Operator at random times