Nfd worker scheduling

hardware_enablement/psap-node-feature-discovery-operator.adoc

@@ -23,6 +23,8 @@ include::modules/creating-nfd-cr-web-console.adoc[leveloffset=+2]
 
 include::modules/psap-configuring-node-feature-discovery.adoc[leveloffset=+1]
 
+include::modules/configuring-nfd-worker-master-pod-scheduling.adoc[leveloffset=+1]
+
 include::modules/nfd-rules-about.adoc[leveloffset=+1]
 
 include::modules/nfd-rules-using.adoc[leveloffset=+1]

modules/configuring-nfd-worker-master-pod-scheduling.adoc

@@ -0,0 +1,114 @@
+// Module included in the following assemblies:
+//
+// * hardware_enablement/psap-node-feature-discovery-operator.adoc
+
+:_mod-docs-content-type: PROCEDURE
+[id="configuring-nfd-worker-master-pod-scheduling_{context}"]
+= Configure node feature discovery worker and controller pod scheduling
+
+[role="_abstract"]
+You can control where Node Feature Discovery (NFD) worker and controller pods are scheduled by configuring node selectors and tolerations in the `NodeFeatureDiscovery` custom resource.
+
+.Prerequisites
+
+* You have access to an {product-title} cluster.
+* You installed the {oc-first}.
+* You logged in as a user with `cluster-admin` privileges.
+* You installed the NFD Operator.
+{empty}
+
+[NOTE]
+====
+If you specify `workerNodeSelector` without the required `workerTolerations`, NFD worker pods are not scheduled on tainted nodes, even when the nodes match the configured node selector.
+====
+
+.Procedure
+
+. Label the target node with the label you will use in `workerNodeSelector`:
++
+[source,terminal]
+----
+$ oc label node <node_name> special-node=true
+----
+
+. Edit the `NodeFeatureDiscovery` custom resource:
++
+[source,terminal]
+----
+$ oc edit NodeFeatureDiscovery nfd-instance -n openshift-nfd
+----
+
+. Configure the scheduling options:
++
+[source,yaml]
+----
+apiVersion: nfd.openshift.io/v1
+kind: NodeFeatureDiscovery
+metadata:
+  name: nfd-instance
+  namespace: openshift-nfd
+spec:
+  enableTaints: false
+  instance: ""
+
+  operand:
+    imagePullPolicy: IfNotPresent
+    servicePort: 12000
+
+    workerNodeSelector:
+      special-node: "true"
+
+    workerTolerations:
+      - key: "node-role.kubernetes.io/ai"
+        operator: "Exists"
+        effect: "NoSchedule"
+
+      - key: "node-role.kubernetes.io/ai"
+        operator: "Exists"
+        effect: "NoExecute"
+
+    masterTolerations:
+      - key: "node-role.kubernetes.io/master"
+        operator: "Exists"
+        effect: "NoSchedule"
+
+      - key: "node-role.kubernetes.io/control-plane"
+        operator: "Exists"
+        effect: "NoSchedule"
+
+  prunerOnDelete: false
+  topologyUpdater: false
+----
++
+where:
++
+`workerNodeSelector`:: Controls where the NFD worker DaemonSet is scheduled. Configure this field under `spec.operand`.
+`workerTolerations`:: Allows the NFD worker pods to tolerate taints applied to the selected nodes. Configure this field under `spec.operand`.
+`masterTolerations`:: Allows NFD controller pods to tolerate taints on control plane nodes. The tolerations must match the node taints for the controller pods to schedule successfully.
+
+.Verification
+
+. Verify that the NFD worker pods are scheduled on the expected nodes:
++
+[source,terminal]
+----
+$ oc get pods -n openshift-nfd -o wide
+----
+
+. Verify the node labels:
++
+[source,terminal]
+----
+$ oc get nodes --show-labels
+----
+
+. Verify the node taints:
++
+[source,terminal]
+----
+$ oc get node <node_name> -o jsonpath='{.spec.taints}'
+----
+
+After verification, NFD worker pods should appear on nodes with the `special-node=true` label, or with the label you configured in `workerNodeSelector`. Controller pods should remain `Running` in the `openshift-nfd` namespace.
+
+