fix release ips when create a container without nictype using older c…

…ni and then upgrade cni and delete

if we create a pod with an older cni version, it won't have a nictype
if we upgrade cni and then delete, we should treat an empty nictype as an infra nictype and
call the invoker delete

cni/network/network.go

@@ -1127,7 +1127,7 @@ func (plugin *NetPlugin) Delete(args *cniSkel.CmdArgs) error {
 			zap.String("endpointID", epInfo.EndpointID))
 		sendEvent(plugin, fmt.Sprintf("Deleting endpoint:%v", epInfo.EndpointID))
 
-		if !nwCfg.MultiTenancy && epInfo.NICType == cns.InfraNIC {
+		if !nwCfg.MultiTenancy && epInfo.NICType != cns.DelegatedVMNIC {
 			// Delegated/secondary nic ips are statically allocated so we don't need to release
 			// Call into IPAM plugin to release the endpoint's addresses.
 			for i := range epInfo.IPAddresses {

cni/network/network_linux.go

@@ -8,7 +8,6 @@ import (
 	"github.com/Azure/azure-container-networking/cns"
 	"github.com/Azure/azure-container-networking/network"
 	"github.com/Azure/azure-container-networking/network/policy"
-	cniSkel "github.com/containernetworking/cni/pkg/skel"
 	cniTypesCurr "github.com/containernetworking/cni/pkg/types/100"
 )
 

network/manager.go

@@ -792,6 +792,8 @@ func cnsEndpointInfotoCNIEpInfos(endpointInfo restserver.EndpointInfo, endpointI
 	return ret
 }
 
+// gets all endpoint infos associated with a container id and populates the network id field
+// nictype may be empty in which case it is likely of type "infra"
 func (nm *networkManager) GetEndpointInfosFromContainerID(containerID string) []*EndpointInfo {
 	ret := []*EndpointInfo{}
 	for _, extIf := range nm.ExternalInterfaces {