Merge e736da5 into 635a275

checkpoint/checkpoint.go

@@ -0,0 +1,76 @@
+package checkpoint
+
+import (
+	"encoding/json"
+	"fmt"
+	"io/ioutil"
+
+	"github.com/intel/multus-cni/types"
+)
+
+const (
+	checkPointfile = "/var/lib/kubelet/device-plugins/kubelet_internal_checkpoint"
+)
+
+type PodDevicesEntry struct {
+	PodUID        string
+	ContainerName string
+	ResourceName  string
+	DeviceIDs     []string
+	AllocResp     []byte
+}
+
+type checkpointData struct {
+	PodDeviceEntries  []PodDevicesEntry
+	RegisteredDevices map[string][]string
+}
+
+type Data struct {
+	Data     checkpointData
+	Checksum uint64
+}
+
+// getPodEntries gets all Pod device allocation entries from checkpoint file
+func getPodEntries() ([]PodDevicesEntry, error) {
+
+	podEntries := []PodDevicesEntry{}
+
+	cpd := &Data{}
+	rawBytes, err := ioutil.ReadFile(checkPointfile)
+	if err != nil {
+		return podEntries, fmt.Errorf("getPodEntries(): error reading file %s\n%v\n", checkPointfile, err)
+
+	}
+
+	if err = json.Unmarshal(rawBytes, cpd); err != nil {
+		return podEntries, fmt.Errorf("getPodEntries(): error unmarshalling raw bytes %v", err)
+	}
+
+	return cpd.Data.PodDeviceEntries, nil
+}
+
+// GetComputeDeviceMap returns a map of resourceName to list of device IDs
+func GetComputeDeviceMap(podID string) (map[string]*types.ResourceInfo, error) {
+
+	resourceMap := make(map[string]*types.ResourceInfo)
+	podEntires, err := getPodEntries()
+
+	if err != nil {
+		return nil, err
+	}
+
+	for _, pod := range podEntires {
+		if pod.PodUID == podID {
+			entry, ok := resourceMap[pod.ResourceName]
+			if ok {
+				// already exists; append to it
+				entry.DeviceIDs = append(entry.DeviceIDs, pod.DeviceIDs...)
+			} else {
+				// new entry
+				resourceMap[pod.ResourceName] = &types.ResourceInfo{DeviceIDs: pod.DeviceIDs}
+			}
+		}
+	}
+
+	return resourceMap, nil
+}

examples/README.md

@@ -60,3 +60,35 @@ A sample `cni-configuration.conf` is provided, typically this file is placed in
 ## Other considerations
 
 Primarily in this setup one thing that one should consider are the aspects of the `macvlan-conf.yml`, which is likely specific to the configuration of the node on which this resides.
+
+## Passing down device information
+Some CNI plugins require specific device information which maybe pre-allocated by K8s device plugin. This could be indicated by providing `k8s.v1.cni.cncf.io/resourceName` annotaton in its network attachment definition CRD. The file [`examples/sriov-net.yaml`](./sriov-net.yaml) shows an example on how to define a Network attachment definition with specific device allocation information. Multus will get allocated device information and make them available for CNI plugin to work on.
+
+In this exmaple (shown below), it is expected that an [SRIOV Device Plugin](https://github.com/intel/sriov-network-device-plugin/) making a pool of SRIOV VFs available to the K8s with `intel.com/sriov` as their resourceName. Any device allocated from this resource pool will be passed down by Multus to the [sriov-cni](https://github.com/intel/sriov-cni/tree/dev/k8s-deviceid-model) plugin in `deviceID` field. This is up to the sriov-cni plugin to capture this information and work with this specific device information.
+
+```yaml
+apiVersion: "k8s.cni.cncf.io/v1"
+kind: NetworkAttachmentDefinition
+metadata:
+  name: sriov-net-a
+  annotations:
+    k8s.v1.cni.cncf.io/resourceName: intel.com/sriov
+spec:
+  config: '{
+  "type": "sriov",
+  "vlan": 1000,
+  "ipam": {
+    "type": "host-local",
+    "subnet": "10.56.217.0/24",
+    "rangeStart": "10.56.217.171",
+    "rangeEnd": "10.56.217.181",
+    "routes": [{
+      "dst": "0.0.0.0/0"
+    }],
+    "gateway": "10.56.217.1"
+  }
+}'
+```
+The [net-resource-sample-pod.yaml](./net-resource-sample-pod.yaml) is an exmaple Pod manifest file that requesting a SRIOV device from a host which is then configured using the above network attachement definition.
+
+>For further information on how to configure SRIOV Device Plugin and SRIOV-CNI please refer to the links given above.

examples/net-resource-sample-pod.yaml

@@ -0,0 +1,21 @@
+apiVersion: v1
+kind: Pod
+metadata:
+  name: testpod1
+  labels:
+    env: test
+  annotations:
+    k8s.v1.cni.cncf.io/networks: sriov-net-a
+spec:
+  containers:
+  - name: appcntr1
+    image: centos/tools
+    imagePullPolicy: IfNotPresent
+    command: [ "/bin/bash", "-c", "--" ]
+    args: [ "while true; do sleep 300000; done;" ]
+    resources:
+      requests:
+        intel.com/sriov: '1'
+      limits:
+        intel.com/sriov: '1'
+  restartPolicy: "Never"

examples/sriov-net.yaml

@@ -0,0 +1,21 @@
+apiVersion: "k8s.cni.cncf.io/v1"
+kind: NetworkAttachmentDefinition
+metadata:
+  name: sriov-net-a
+  annotations:
+    k8s.v1.cni.cncf.io/resourceName: intel.com/sriov
+spec:
+  config: '{
+  "type": "sriov",
+  "vlan": 1000,
+  "ipam": {
+    "type": "host-local",
+    "subnet": "10.56.217.0/24",
+    "rangeStart": "10.56.217.171",
+    "rangeEnd": "10.56.217.181",
+    "routes": [{
+      "dst": "0.0.0.0/0"
+    }],
+    "gateway": "10.56.217.1"
+  }
+}'

k8sclient/k8sclient.go

@@ -31,10 +31,15 @@ import (
 	"github.com/containernetworking/cni/libcni"
 	"github.com/containernetworking/cni/pkg/skel"
 	cnitypes "github.com/containernetworking/cni/pkg/types"
+	"github.com/intel/multus-cni/checkpoint"
 	"github.com/intel/multus-cni/logging"
 	"github.com/intel/multus-cni/types"
 )
 
+const (
+	resourceNameAnnot = "k8s.v1.cni.cncf.io/resourceName"
+)
+
 // NoK8sNetworkError indicates error, no network in kubernetes
 type NoK8sNetworkError struct {
 	message string
@@ -131,16 +136,16 @@ func setPodNetworkAnnotation(client KubeClient, namespace string, pod *v1.Pod, n
 	return pod, nil
 }
 
-func getPodNetworkAnnotation(client KubeClient, k8sArgs *types.K8sArgs) (string, string, error) {
+func getPodNetworkAnnotation(client KubeClient, k8sArgs *types.K8sArgs) (string, string, string, error) {
 	var err error
 
 	logging.Debugf("getPodNetworkAnnotation: %v, %v", client, k8sArgs)
 	pod, err := client.GetPod(string(k8sArgs.K8S_POD_NAMESPACE), string(k8sArgs.K8S_POD_NAME))
 	if err != nil {
-		return "", "", logging.Errorf("getPodNetworkAnnotation: failed to query the pod %v in out of cluster comm: %v", string(k8sArgs.K8S_POD_NAME), err)
+		return "", "", "", logging.Errorf("getPodNetworkAnnotation: failed to query the pod %v in out of cluster comm: %v", string(k8sArgs.K8S_POD_NAME), err)
 	}
 
-	return pod.Annotations["k8s.v1.cni.cncf.io/networks"], pod.ObjectMeta.Namespace, nil
+	return pod.Annotations["k8s.v1.cni.cncf.io/networks"], pod.ObjectMeta.Namespace, string(pod.UID), nil
 }
 
 func parsePodNetworkObjectName(podnetwork string) (string, string, string, error) {
@@ -326,7 +331,8 @@ func cniConfigFromNetworkResource(customResource *types.NetworkAttachmentDefinit
 	return config, nil
 }
 
-func getKubernetesDelegate(client KubeClient, net *types.NetworkSelectionElement, confdir string) (*types.DelegateNetConf, error) {
+func getKubernetesDelegate(client KubeClient, net *types.NetworkSelectionElement, confdir string, resourceMap map[string]*types.ResourceInfo) (*types.DelegateNetConf, error) {
+
 	logging.Debugf("getKubernetesDelegate: %v, %v, %s", client, net, confdir)
 	rawPath := fmt.Sprintf("/apis/k8s.cni.cncf.io/v1/namespaces/%s/network-attachment-definitions/%s", net.Namespace, net.Name)
 	netData, err := client.GetRawWithPath(rawPath)
@@ -339,12 +345,26 @@ func getKubernetesDelegate(client KubeClient, net *types.NetworkSelectionElement
 		return nil, logging.Errorf("getKubernetesDelegate: failed to get the netplugin data: %v", err)
 	}
 
+	// Get resourceName annotation from NetDefinition
+	deviceID := ""
+	resourceName, ok := customResource.Metadata.Annotations[resourceNameAnnot]
+	if ok {
+		// ResourceName annotation is found; try to get device info from resourceMap
+		entry, ok := resourceMap[resourceName]
+		if ok {
+			if idCount := len(entry.DeviceIDs); idCount > 0 && idCount > entry.Index {
+				deviceID = entry.DeviceIDs[entry.Index]
+				entry.Index++ // increment Index for next delegate
+			}
+		}
+	}
+
 	configBytes, err := cniConfigFromNetworkResource(customResource, confdir)
 	if err != nil {
 		return nil, err
 	}
 
-	delegate, err := types.LoadDelegateNetConf(configBytes, net.InterfaceRequest)
+	delegate, err := types.LoadDelegateNetConf(configBytes, net.InterfaceRequest, deviceID)
 	if err != nil {
 		return nil, err
 	}
@@ -447,11 +467,17 @@ func GetK8sClient(kubeconfig string, kubeClient KubeClient) (KubeClient, error)
 func GetK8sNetwork(k8sclient KubeClient, k8sArgs *types.K8sArgs, confdir string) ([]*types.DelegateNetConf, error) {
 	logging.Debugf("GetK8sNetwork: %v, %v, %v", k8sclient, k8sArgs, confdir)
 
-	netAnnot, defaultNamespace, err := getPodNetworkAnnotation(k8sclient, k8sArgs)
+	netAnnot, defaultNamespace, podID, err := getPodNetworkAnnotation(k8sclient, k8sArgs)
 	if err != nil {
 		return nil, err
 	}
 
+	// Get Pod ComputeDevices info
+	resourceMap, err := checkpoint.GetComputeDeviceMap(podID)
+	if err != nil {
+		return nil, logging.Errorf("GetK8sNetwork: failed to get resourceMap for PodUID: %v %v", podID, err)
+	}
+
 	if len(netAnnot) == 0 {
 		return nil, &NoK8sNetworkError{"no kubernetes network found"}
 	}
@@ -464,7 +490,7 @@ func GetK8sNetwork(k8sclient KubeClient, k8sArgs *types.K8sArgs, confdir string)
 	// Read all network objects referenced by 'networks'
 	var delegates []*types.DelegateNetConf
 	for _, net := range networks {
-		delegate, err := getKubernetesDelegate(k8sclient, net, confdir)
+		delegate, err := getKubernetesDelegate(k8sclient, net, confdir, resourceMap)
 		if err != nil {
 			return nil, logging.Errorf("GetK8sNetwork: failed getting the delegate: %v", err)
 		}

types/conf.go

@@ -50,7 +50,16 @@ func LoadDelegateNetConfList(bytes []byte, delegateConf *DelegateNetConf) error
 }
 
 // Convert raw CNI JSON into a DelegateNetConf structure
-func LoadDelegateNetConf(bytes []byte, ifnameRequest string) (*DelegateNetConf, error) {
+func LoadDelegateNetConf(bytes []byte, ifnameRequest, deviceID string) (*DelegateNetConf, error) {
+	// If deviceID is present, inject this into delegate config
+	if deviceID != "" {
+		if updatedBytes, err := delegateAddDeviceID(bytes, deviceID); err != nil {
+			return nil, logging.Errorf("error in LoadDelegateNetConf - delegateAddDeviceID unable to update delegate config: %v", err)
+		} else {
+			bytes = updatedBytes
+		}
+	}
+
 	delegateConf := &DelegateNetConf{}
 	logging.Debugf("LoadDelegateNetConf: %s, %s", string(bytes), ifnameRequest)
 	if err := json.Unmarshal(bytes, &delegateConf.Conf); err != nil {
@@ -190,7 +199,7 @@ func LoadNetConf(bytes []byte) (*NetConf, error) {
 		if err != nil {
 			return nil, logging.Errorf("error marshalling delegate %d config: %v", idx, err)
 		}
-		delegateConf, err := LoadDelegateNetConf(bytes, "")
+		delegateConf, err := LoadDelegateNetConf(bytes, "", "")
 		if err != nil {
 			return nil, logging.Errorf("failed to load delegate %d config: %v", idx, err)
 		}
@@ -210,3 +219,21 @@ func (n *NetConf) AddDelegates(newDelegates []*DelegateNetConf) error {
 	n.Delegates = append(n.Delegates, newDelegates...)
 	return nil
 }
+
+// delegateAddDeviceID injects deviceID information in delegate bytes
+func delegateAddDeviceID(inBytes []byte, deviceID string) ([]byte, error) {
+	var rawConfig map[string]interface{}
+	var err error
+
+	err = json.Unmarshal(inBytes, &rawConfig)
+	if err != nil {
+		return nil, logging.Errorf("delegateAddDeviceID: failed to unmarshal inBytes: %v", err)
+	}
+	// Inject deviceID
+	rawConfig["deviceID"] = deviceID
+	configBytes, err := json.Marshal(rawConfig)
+	if err != nil {
+		return nil, logging.Errorf("delegateAddDeviceID: failed to re-marshal Spec.Config: %v", err)
+	}
+	return configBytes, nil
+}

types/types.go

@@ -119,3 +119,9 @@ type K8sArgs struct {
 	K8S_POD_NAMESPACE          types.UnmarshallableString
 	K8S_POD_INFRA_CONTAINER_ID types.UnmarshallableString
 }
+
+// ResourceInfo is struct to hold Pod device allocation information
+type ResourceInfo struct {
+	Index     int
+	DeviceIDs []string
+}