fixes

Signed-off-by: Miek Gieben <miek@miek.nl>

plugin/traffic/README.md

@@ -23,7 +23,7 @@ endpoints need to be drained from it.
 every 10 seconds. The plugin hands out responses that adhere to these assignments. Each DNS response
 contains a single IP address that's considered the best one. *Traffic* will load balance A and AAAA
 queries. The TTL on these answer is set to 5s. It will only return successful responses either with
-an answer or otherwise a NODATA response. NXDOMAIN responses will *never* be sent.
+an answer or otherwise a NODATA response. Queries for non-existent clusters get a NXDOMAIN.
 
 The *traffic* plugin has no notion of draining, drop overload and anything that advanced, *it just
 acts upon assignments*. This is means that if a endpoint goes down and *traffic* has not seen a new
@@ -35,8 +35,9 @@ assignment yet, it will still include this endpoint address in responses.
 traffic TO...
 ~~~
 
- *  **TO...** are the Envoy control plane endpoint to connect to. The syntax mimics the *forward*
+This enabled the *traffic* plugin, with a default node id of `coredns` and no TLS.
-    plugin and must start with `grpc://`.
+
+*  **TO...** are the Envoy control plane endpoint to connect to. This must start with `grpc://`.
 
 The extended syntax is available is you want more control.
 
@@ -46,7 +47,7 @@ traffic TO... {
     node ID
     tls CERT KEY CA
     tls_servername NAME
- }
+}
 ~~~
 
 *  node **ID** is how *traffic* identifies itself to the control plane. This defaults to `coredns`.
@@ -57,7 +58,7 @@ traffic TO... {
   * `tls` **CA** - no client authentication is used, and the file CA is used to verify the server certificate
   * `tls` **CERT** **KEY** - client authentication is used with the specified cert/key pair.
     The server certificate is verified with the system CAs.
-  * `tls` **CERT** **KEY**  **CA** - client authentication is used with the specified cert/key pair.
+  * `tls` **CERT** **KEY** **CA** - client authentication is used with the specified cert/key pair.
     The server certificate is verified using the specified CA file.
 
 * `tls_servername` **NAME** allows you to set a server name in the TLS configuration. This is needed
@@ -120,9 +121,9 @@ Multiple **TO** addresses is not implemented.
 
 ## TODO
 
-* reconnecting the stream
 * acking responses
 * correctly tracking versions and pruning old clusters.
 * metrics?
+* how to exactly deal with health status from the endpoints.
 * testing
-* credentials (other than TLS)
+* credentials (other than TLS) - how/what?

plugin/traffic/traffic.go

@@ -39,18 +39,20 @@ func (t *Traffic) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg
 		return plugin.NextOrFailure(t.Name(), t.Next, ctx, w, r)
 	}
 
-	addr := t.c.Select(cluster)
-	if addr != nil {
-		log.Debugf("Found endpoint %q for %q", addr, cluster)
-	} else {
-		log.Debugf("No healthy endpoints found for %q", cluster)
-	}
-
 	m := new(dns.Msg)
 	m.SetReply(r)
 	m.Authoritative = true
 
+	addr, ok := t.c.Select(cluster)
+	if !ok {
+		m.Ns = soa(state.Zone)
+		m.Rcode = dns.RcodeNameError
+		w.WriteMsg(m)
+		return 0, nil
+	}
+
 	if addr == nil {
+		log.Debugf("No (healthy) endpoints found for %q", cluster)
 		m.Ns = soa(state.Zone)
 		w.WriteMsg(m)
 		return 0, nil

plugin/traffic/xds/assignment.go

@@ -54,10 +54,10 @@ func (a *assignment) clusters() []string {
 
 // Select selects a backend from cla, using weighted random selection. It only selects
 // backends that are reporting healthy.
-func (a *assignment) Select(cluster string) net.IP {
+func (a *assignment) Select(cluster string) (net.IP, bool) {
 	cla := a.clusterLoadAssignment(cluster)
 	if cla == nil {
-		return nil
+		return nil, false
 	}
 
 	total := 0
@@ -81,7 +81,7 @@ func (a *assignment) Select(cluster string) net.IP {
 				//					continue
 				//				}
 				if r == i {
-					return net.ParseIP(lb.GetEndpoint().GetAddress().GetSocketAddress().GetAddress())
+					return net.ParseIP(lb.GetEndpoint().GetAddress().GetSocketAddress().GetAddress()), true
 				}
 				i++
 			}
@@ -98,9 +98,9 @@ func (a *assignment) Select(cluster string) net.IP {
 			//			}
 			r -= int(lb.GetLoadBalancingWeight().GetValue())
 			if r <= 0 {
-				return net.ParseIP(lb.GetEndpoint().GetAddress().GetSocketAddress().GetAddress())
+				return net.ParseIP(lb.GetEndpoint().GetAddress().GetSocketAddress().GetAddress()), true
 			}
 		}
 	}
-	return nil
+	return nil, false
 }

plugin/traffic/xds/client.go

@@ -216,5 +216,6 @@ func (c *Client) Receive(stream adsStream) error {
 	}
 }
 
-// Select returns an address that is deemed to be the correct one for this cluster.
+// Select returns an address that is deemed to be the correct one for this cluster. The returned
-func (c *Client) Select(cluster string) net.IP { return c.assignments.Select(cluster) }
+// boolean indicates if the cluster exists.
+func (c *Client) Select(cluster string) (net.IP, bool) { return c.assignments.Select(cluster) }