[core] improved error reporting during env deployment/configuration

core/controlcommands/commandqueue.go

@@ -101,7 +101,7 @@ func (m *CommandQueue) Start() {
 				}
 				if err == nil && response == nil {
 					log.WithField("partition", entry.cmd.GetEnvironmentId().String()).
-						Error("nil response")
+						Errorf("did not receive neither response nor error for %s", entry.cmd.GetName())
 				}
 
 				entry.callback <- response
@@ -198,6 +198,11 @@ func (m *CommandQueue) commit(command MesosCommand) (response MesosCommandRespon
 	// Wait for goroutines to finish
 	for i := 0; i < len(command.targets()); i++ {
 		respSemaphore := <-semaphore
+		// for the sake of better error propagation, we treat a lack of response as a response with error,
+		// even though it's not technically the same. it can be surely done better, but it would require a larger refactoring.
+		if respSemaphore.err != nil && respSemaphore.response == nil {
+			respSemaphore.response = NewMesosCommandResponse(command, respSemaphore.err)
+		}
 		responses[respSemaphore.receiver] = respSemaphore.response
 		if respSemaphore.err != nil {
 			sendErrorList = append(sendErrorList, respSemaphore.err)
@@ -215,12 +220,11 @@ func (m *CommandQueue) commit(command MesosCommand) (response MesosCommandRespon
 			}
 			return
 		}(), "\n"))
-		return
 	}
 	response = consolidateResponses(command, responses)
 
 	log.WithField("partition", command.GetEnvironmentId().String()).
 		Debug("responses consolidated, CommandQueue commit done")
 
-	return response, nil
+	return response, err
 }

core/controlcommands/mesoscommandservent.go

@@ -144,7 +144,7 @@ func (s *Servent) RunCommand(cmd MesosCommand, receiver MesosCommandTarget) (Mes
 		// By the time we get here, ProcessResponse should have already added a Response to the
 		// pending call, and removed it from servent.pending.
 	case <-time.After(cmd.GetResponseTimeout()):
-		call.Error = fmt.Errorf("MesosCommand %s timed out for task %s", cmd.GetName(), receiver.TaskId.Value)
+		call.Error = fmt.Errorf("%s timed out for task %s", cmd.GetName(), receiver.TaskId.Value)
 
 		log.WithPrefix("servent").
 			WithField("partition", cmd.GetEnvironmentId().String()).

core/task/manager.go

@@ -754,7 +754,7 @@ func (m *Manager) configureTasks(envId uid.ID, tasks Tasks) error {
 	close(notify)
 
 	if response == nil {
-		return errors.New("nil response")
+		return fmt.Errorf("no response from Mesos to CONFIGURE transition request within %ds timeout", int(cmd.ResponseTimeout.Seconds()))
 	}
 
 	if response.IsMultiResponse() {
@@ -765,14 +765,7 @@ func (m *Manager) configureTasks(envId uid.ID, tasks Tasks) error {
 			task := m.GetTask(k.TaskId.Value)
 			var taskDescription string
 			if task != nil {
-				tci := task.GetTaskCommandInfo()
-				tciValue := "unknown command"
-				if tci.Value != nil {
-					tciValue = *tci.Value
-				}
-
-				taskDescription = fmt.Sprintf("task '%s' on %s (id %s, name %s) failed with error: %s", tciValue, task.GetHostname(), task.GetTaskId(), task.GetName(), v.Error())
-
+				taskDescription = fmt.Sprintf("task '%s' on %s (id %s) failed with error: %s", task.GetParent().GetName(), task.GetHostname(), task.GetTaskId(), v.Error())
 			} else {
 				taskDescription = fmt.Sprintf("unknown task (id %s) failed with error: %s", k.TaskId.Value, v.Error())
 			}

core/task/task.go

@@ -76,6 +76,7 @@ type parentRole interface {
 	ConsolidatedVarStack() (varStack map[string]string, err error)
 	CollectInboundChannels() []channel.Inbound
 	SendEvent(event.Event)
+	GetName() string
 }
 
 type Traits struct {