WIP: Change GPU plugin NFD init container to run-time container

To work around kustomize inability to enforce correct init container, and because this is more likely way how it will work once NFD drops support for hooks: kubernetes-sigs/node-feature-discovery#856 Signed-off-by: Eero Tamminen <[email protected]>
intel · Aug 29, 2022 · fa7cfa7 · fa7cfa7
1 parent e1d6928
commit fa7cfa7
Show file tree

Hide file tree

Showing 4 changed files with 45 additions and 25 deletions.
diff --git a/deployments/gpu_plugin/overlays/fake_devices/add-nfd-feature-file.yaml b/deployments/gpu_plugin/overlays/fake_devices/add-nfd-feature-file.yaml
@@ -0,0 +1,34 @@
+apiVersion: apps/v1
+kind: DaemonSet
+metadata:
+  name: intel-gpu-plugin
+spec:
+  template:
+    spec:
+      containers:
+      - name: intel-gpu-nfd
+        # convert generated sysfs content to NFD feature labels file
+        image: intel/intel-gpu-initcontainer:devel
+        imagePullPolicy: IfNotPresent
+        securityContext:
+          readOnlyRootFilesystem: true
+          allowPrivilegeEscalation: false
+          capabilities:
+            drop: [ "ALL" ]
+        volumeMounts:
+        - name: sysfs
+          # expects sysfs here
+          mountPath: /host-sys
+          readOnly: true
+        - name: nfd-features
+          mountPath: /nfd
+          readOnly: false
+        workingDir: /usr/local/bin/gpu-sw
+        # needed until GPU plugin drops NFD hook usage due to:
+        # https://github.com/kubernetes-sigs/node-feature-discovery/issues/856
+        command: ["sh", "-c", "while true; do ./intel-gpu-nfdhook | tee /nfd/fake-gpu; sleep 99999; done"]
+      volumes:
+      - name: nfd-features
+        hostPath:
+          path: /etc/kubernetes/node-feature-discovery/features.d/
+          type: DirectoryOrCreate
diff --git a/deployments/gpu_plugin/overlays/fake_devices/del-intel-gpu-initcontainer.yaml b/deployments/gpu_plugin/overlays/fake_devices/del-intel-gpu-initcontainer.yaml
@@ -0,0 +1,6 @@
+$patch: delete
+spec:
+  template:
+    spec:
+      initContainers:
+      - name: intel-gpu-initcontainer
diff --git a/deployments/gpu_plugin/overlays/fake_devices/kustomization.yaml b/deployments/gpu_plugin/overlays/fake_devices/kustomization.yaml
@@ -6,9 +6,11 @@ configMapGenerator:
 - name: fakedev-config
   files:
   - fakedev-config.json
+patchesStrategicMerge:
+  # NFD feature file change is obsolete after GPU plugin moves away from NFD hooks
+- del-intel-gpu-initcontainer.yaml
 patches:
   - fake-device-volumes.yaml
   - generate-fake-devices.yaml
-  # NFD feature file change is obsolete after GPU plugin moves away from NFD hooks:
-  # https://github.com/kubernetes-sigs/node-feature-discovery/issues/856
-  - use-nfd-feature-file.yaml
+  # NFD feature file change is obsolete after GPU plugin moves away from NFD hooks
+  - add-nfd-feature-file.yaml
diff --git a/deployments/gpu_plugin/overlays/fake_devices/use-nfd-feature-file.yaml b/deployments/gpu_plugin/overlays/fake_devices/use-nfd-feature-file.yaml