Fixed an endpoint delete issue with a possible stale sbox

Stale sandbox could be a result of stress tests on ungraceful
restarts such as #17984

Signed-off-by: Madhu Venugopal <madhu@docker.com>

hack/vendor.sh

@@ -22,7 +22,7 @@ clone git github.com/vdemeester/shakers 3c10293ce22b900c27acad7b28656196fcc2f73b
 clone git golang.org/x/net 47990a1ba55743e6ef1affd3a14e5bac8553615d https://github.com/golang/net.git
 
 #get libnetwork packages
-clone git github.com/docker/libnetwork e8ebc0bf6510343c88d162db08b3d855cbbe75b9
+clone git github.com/docker/libnetwork b4ddf18317b19d6e4bcc821145589749206a7d00
 clone git github.com/armon/go-metrics eb0af217e5e9747e41dd5303755356b62d28e3ec
 clone git github.com/hashicorp/go-msgpack 71c2886f5a673a35f909803f38ece5810165097b
 clone git github.com/hashicorp/memberlist 9a1e242e454d2443df330bdd51a436d5a9058fc4

vendor/src/github.com/docker/libnetwork/endpoint.go

@@ -584,7 +584,8 @@ func (ep *endpoint) Delete() error {
 	ep.Lock()
 	epid := ep.id
 	name := ep.name
-	if ep.sandboxID != "" {
+	sb, _ := n.getController().SandboxByID(ep.sandboxID)
+	if sb != nil {
 		ep.Unlock()
 		return &ActiveContainerError{name: name, id: epid}
 	}

vendor/src/github.com/docker/libnetwork/store.go

@@ -176,7 +176,6 @@ func (n *network) getEndpointsFromStore() ([]*endpoint, error) {
 
 		for _, kvo := range kvol {
 			ep := kvo.(*endpoint)
-			ep.network = n
 			epl = append(epl, ep)
 		}
 	}