fn: error handling updates (#1337)

* docker-pull timeout is now a 504 which classifies it as a
service error. Avoid using 503 to make sure LB does not retry.

* Only applicable to detached mode, a timeout on LB is
now a ErrServiceReservationFailure (500). In detached mode,
this is unlikely to make it back to a client and it is mostly
for documentation/metrics purposes.
  
* For Triggers, avoid scrubbing service code.

api/agent/lb_agent.go

@@ -304,6 +304,18 @@ func (a *lbAgent) handleCallEnd(ctx context.Context, call *call, err error, isFo
 		if err == nil {
 			statsComplete(ctx)
 			recordCallLatency(ctx, call, completedMetricName)
+		} else if err == context.DeadlineExceeded {
+			// We are here because we were unable to service this request for the given
+			// reservation. In detached case, the reservation is calculated based on estimated
+			// total time to run a request. (See: spawnPlaceCall) Otherwise, there's no set
+			// deadline in the request context. This is also a bit more robust going forward
+			// if we start enforcing a maximum overall deadline for clients. For detached case, the
+			// error is unlikely to be delivered to the client since this is essentially an async
+			// operation.
+			statsTimedout(ctx)
+			recordCallLatency(ctx, call, timedoutMetricName)
+			// We have failed: http 500 Internal Server Error
+			return models.ErrServiceReservationFailure
 		}
 	} else {
 		statsDequeue(ctx)
@@ -318,10 +330,6 @@ func (a *lbAgent) handleCallEnd(ctx context.Context, call *call, err error, isFo
 		statsTooBusy(ctx)
 		recordCallLatency(ctx, call, serverBusyMetricName)
 		return models.ErrCallTimeoutServerBusy
-	} else if err == context.DeadlineExceeded {
-		statsTimedout(ctx)
-		recordCallLatency(ctx, call, timedoutMetricName)
-		return models.ErrCallTimeout
 	} else if err == context.Canceled {
 		statsCanceled(ctx)
 		recordCallLatency(ctx, call, canceledMetricName)

api/models/error.go

@@ -37,7 +37,7 @@ var (
 		error: errors.New("Timed out - server too busy"),
 	}
 	ErrDockerPullTimeout = err{
-		code:  http.StatusServiceUnavailable,
+		code:  http.StatusGatewayTimeout,
 		error: errors.New("Docker pull timed out"),
 	}
 	ErrContainerInitTimeout = err{
@@ -188,7 +188,10 @@ var (
 		code:  http.StatusInternalServerError,
 		error: errors.New("Unable to find the call handle"),
 	}
-
+	ErrServiceReservationFailure = err{
+		code:  http.StatusInternalServerError,
+		error: errors.New("Unable to service the request for the reservation period"),
+	}
 	ErrContainerInitFail = err{
 		code:  http.StatusBadGateway,
 		error: errors.New("container failed to initialize, please ensure you are using the latest fdk / format and check the logs"),

api/server/runner_httptrigger.go

@@ -76,13 +76,13 @@ func (trw *triggerResponseWriter) Write(b []byte) (int, error) {
 	return trw.inner.Write(b)
 }
 
-func (trw *triggerResponseWriter) WriteHeader(statusCode int) {
+func (trw *triggerResponseWriter) WriteHeader(serviceStatus int) {
 	if trw.committed {
 		return
 	}
 	trw.committed = true
 
-	var fnStatus int
+	userStatus := 0
 	realHeaders := trw.Header()
 	gwHeaders := make(http.Header, len(realHeaders))
 	for k, vs := range realHeaders {
@@ -96,7 +96,7 @@ func (trw *triggerResponseWriter) WriteHeader(statusCode int) {
 			if len(vs) > 0 {
 				statusInt, err := strconv.Atoi(vs[0])
 				if err == nil {
-					fnStatus = statusInt
+					userStatus = statusInt
 				}
 			}
 		case k == "Content-Type", k == "Fn-Call-Id":
@@ -113,14 +113,14 @@ func (trw *triggerResponseWriter) WriteHeader(statusCode int) {
 	}
 
 	// XXX(reed): simplify / add tests for these behaviors...
-	gatewayStatus := 200
+	finalStatus := 200
-	if statusCode >= 400 {
+	if serviceStatus >= 400 {
-		gatewayStatus = 502
+		finalStatus = serviceStatus
-	} else if fnStatus > 0 {
+	} else if userStatus > 0 {
-		gatewayStatus = fnStatus
+		finalStatus = userStatus
 	}
 
-	trw.inner.WriteHeader(gatewayStatus)
+	trw.inner.WriteHeader(finalStatus)
 }
 
 var skipTriggerHeaders = map[string]bool{