[REDHAT ISSUE] Make init container images configurable through helm-charts (#912)

Author: nikhilsk

api/v1alpha1/deviceconfig_types.go

@@ -773,7 +773,7 @@ type UtilsContainerSpec struct {
 
 // CommonConfigSpec contains the common config across operator and operands
 type CommonConfigSpec struct {
-	// InitContainerImage is being used for the operands pods, i.e. metrics exporter, test runner, device plugin and node labeller
+	// InitContainerImage is being used for the operands pods, i.e. metrics exporter, test runner, device plugin, device config manager and node labeller
 	//+operator-sdk:csv:customresourcedefinitions:type=spec,displayName="InitContainerImage",xDescriptors={"urn:alm:descriptor:com.amd.deviceconfigs:initContainerImage"}
 	// +optional
 	InitContainerImage string `json:"initContainerImage,omitempty"`

bundle/manifests/amd-gpu-operator.clusterserviceversion.yaml

@@ -83,7 +83,8 @@ spec:
         x-descriptors:
         - urn:alm:descriptor:com.amd.deviceconfigs:commonConfig
       - description: InitContainerImage is being used for the operands pods, i.e.
-          metrics exporter, test runner, device plugin and node labeller
+          metrics exporter, test runner, device plugin, device config manager and
+          node labeller
         displayName: InitContainerImage
         path: commonConfig.initContainerImage
         x-descriptors:

bundle/manifests/amd.com_deviceconfigs.yaml

@@ -51,8 +51,8 @@ spec:
                 properties:
                   initContainerImage:
                     description: InitContainerImage is being used for the operands
-                      pods, i.e. metrics exporter, test runner, device plugin and
-                      node labeller
+                      pods, i.e. metrics exporter, test runner, device plugin, device
+                      config manager and node labeller
                     type: string
                   utilsContainer:
                     description: UtilsContainer contains parameters to configure operator's

config/crd/bases/amd.com_deviceconfigs.yaml

@@ -47,8 +47,8 @@ spec:
                 properties:
                   initContainerImage:
                     description: InitContainerImage is being used for the operands
-                      pods, i.e. metrics exporter, test runner, device plugin and
-                      node labeller
+                      pods, i.e. metrics exporter, test runner, device plugin, device
+                      config manager and node labeller
                     type: string
                   utilsContainer:
                     description: UtilsContainer contains parameters to configure operator's

config/manifests/bases/amd-gpu-operator.clusterserviceversion.yaml

@@ -54,7 +54,8 @@ spec:
         x-descriptors:
         - urn:alm:descriptor:com.amd.deviceconfigs:commonConfig
       - description: InitContainerImage is being used for the operands pods, i.e.
-          metrics exporter, test runner, device plugin and node labeller
+          metrics exporter, test runner, device plugin, device config manager and
+          node labeller
         displayName: InitContainerImage
         path: commonConfig.initContainerImage
         x-descriptors:

helm-charts-k8s/crds/deviceconfig-crd.yaml

@@ -56,7 +56,8 @@ spec:
                 properties:
                   initContainerImage:
                     description: InitContainerImage is being used for the operands pods,
-                      i.e. metrics exporter, test runner, device plugin and node labeller
+                      i.e. metrics exporter, test runner, device plugin, device config
+                      manager and node labeller
                     type: string
                   utilsContainer:
                     description: UtilsContainer contains parameters to configure operator's

helm-charts-openshift/crds/deviceconfig-crd.yaml

@@ -56,7 +56,8 @@ spec:
                 properties:
                   initContainerImage:
                     description: InitContainerImage is being used for the operands pods,
-                      i.e. metrics exporter, test runner, device plugin and node labeller
+                      i.e. metrics exporter, test runner, device plugin, device config
+                      manager and node labeller
                     type: string
                   utilsContainer:
                     description: UtilsContainer contains parameters to configure operator's

internal/configmanager/configmanager.go

@@ -54,6 +54,7 @@ const (
 	defaultConfigManagerImage = "docker.io/rocm/device-config-manager:v1.4.0"
 	ConfigManagerName         = "device-config-manager"
 	defaultSAName             = "amd-gpu-operator-config-manager"
+	defaultInitContainerImage = "busybox:1.36"
 )
 
 var configManagerLabelPair = []string{"app.kubernetes.io/name", ConfigManagerName}
@@ -289,6 +290,10 @@ func (nl *configManager) SetConfigManagerAsDesired(ds *appsv1.DaemonSet, devConf
 
 	serviceaccount := defaultSAName
 	gracePeriod := int64(1)
+	initContainerImage := defaultInitContainerImage
+	if devConfig.Spec.CommonConfig.InitContainerImage != "" {
+		initContainerImage = devConfig.Spec.CommonConfig.InitContainerImage
+	}
 	initContainerCommand := "if [ \"$SIM_ENABLE\" = \"true\" ]; then exit 0; fi; while [ ! -d /host-sys/class/kfd ] || [ ! -d /host-sys/module/amdgpu/drivers/ ]; do echo \"amdgpu driver is not loaded \"; sleep 2 ;done"
 	switch devConfig.Spec.Driver.DriverType {
 	case utils.DriverTypeVFPassthrough:
@@ -307,7 +312,7 @@ func (nl *configManager) SetConfigManagerAsDesired(ds *appsv1.DaemonSet, devConf
 				InitContainers: []v1.Container{
 					{
 						Name:            "driver-init",
-						Image:           "busybox:1.36",
+						Image:           initContainerImage,
 						Command:         []string{"sh", "-c", initContainerCommand},
 						SecurityContext: &v1.SecurityContext{Privileged: ptr.To(true)},
 						VolumeMounts: []v1.VolumeMount{