Remove dispatch metrics

CHANGELOG.md

@@ -1,3 +1,7 @@
+28.0.0
+------
+- Internal pipeline refactor removes two internal metrics
+
 27.0.0
 ------
 - Simplify internal tracking of the source address of a metric.  The current rules are:

METRICS.md

@@ -16,7 +16,6 @@ Metrics:
 
 | Name                                        | type                | tags                         | description
 | ------------------------------------------- | ------------------- | ---------------------------- | -----------
-| aggregator.metrics_received                 | gauge (flush)       | aggregator_id                | The number of datapoints received during the flush interval
 | aggregator.metricmaps_received              | gauge (flush)       | aggregator_id                | The number of datapoint batches received during the flush interval
 | aggregator.aggregation_time                 | gauge (time)        | aggregator_id                | The time taken (in ms) to aggregate all counter and timer
 |                                             |                     |                              | datapoints in this flush interval
@@ -84,7 +83,6 @@ the samples are reset.
 
 | Channel name              | Additional tags | Description
 | ------------------------- | --------------- | -----------
-| dispatch_aggregator_batch | aggregator_id   | Channel to dispatch metrics to a specific aggregator.
 | dispatch_aggregator_map   | aggregator_id   | Channel to dispatch metric maps to a given aggregator.
 | backend_events_sem        |                 | Semaphore limiting the number of events in flight at once.  Corresponds to
 |                           |                 | the `--max-concurrent-events` flag.

events.go

@@ -92,5 +92,10 @@ type Event struct {
 	AlertType AlertType
 }
 
+func (e *Event) AddTagsSetSource(additionalTags Tags, newSource Source) {
+	e.Tags = e.Tags.Concat(additionalTags)
+	e.Source = newSource
+}
+
 // Events represents a list of events.
-type Events []Event
+type Events []*Event

fixtures_test.go

@@ -2,7 +2,6 @@ package gostatsd
 
 import (
 	"context"
-	"sync"
 	"testing"
 	"time"
 
@@ -24,44 +23,3 @@ func testContext(t *testing.T) (context.Context, func()) {
 	}()
 	return ctxTest, completeTest
 }
-
-type capturingHandler struct {
-	mu sync.Mutex
-	m  []*Metric
-	e  []*Event
-}
-
-func (ch *capturingHandler) EstimatedTags() int {
-	return 0
-}
-
-func (ch *capturingHandler) DispatchMetrics(ctx context.Context, metrics []*Metric) {
-	ch.mu.Lock()
-	defer ch.mu.Unlock()
-	for _, m := range metrics {
-		m.DoneFunc = nil // Clear DoneFunc because it contains non-predictable variable data which interferes with the tests
-		ch.m = append(ch.m, m)
-	}
-}
-
-// DispatchMetricMap re-dispatches a metric map through capturingHandler.DispatchMetrics
-func (ch *capturingHandler) DispatchMetricMap(ctx context.Context, mm *MetricMap) {
-	mm.DispatchMetrics(ctx, ch)
-}
-
-func (ch *capturingHandler) DispatchEvent(ctx context.Context, e *Event) {
-	ch.mu.Lock()
-	defer ch.mu.Unlock()
-	ch.e = append(ch.e, e)
-}
-
-func (ch *capturingHandler) WaitForEvents() {
-}
-
-func (ch *capturingHandler) GetMetrics() []*Metric {
-	ch.mu.Lock()
-	defer ch.mu.Unlock()
-	m := make([]*Metric, len(ch.m))
-	copy(m, ch.m)
-	return m
-}

internal/fixtures/metrics.go

@@ -0,0 +1,80 @@
+package fixtures
+
+import (
+	"fmt"
+
+	"github.com/atlassian/gostatsd"
+)
+
+type MetricOpt func(m *gostatsd.Metric)
+
+// MakeMetric provides a way to build a metric for tests.  Hopefully over
+// time this will be used more, bringing more consistency to tests.
+func MakeMetric(opts ...MetricOpt) *gostatsd.Metric {
+	m := &gostatsd.Metric{
+		Type: gostatsd.COUNTER,
+		Name: "name",
+		Rate: 1,
+		Tags: gostatsd.Tags{
+			"foo:bar",
+			"host:baz",
+		},
+		Source: "baz",
+	}
+	for _, opt := range opts {
+		opt(m)
+	}
+	return m
+}
+
+func Name(n string) MetricOpt {
+	return func(m *gostatsd.Metric) {
+		m.Name = n
+	}
+}
+
+func AddTag(t ...string) MetricOpt {
+	return func(m *gostatsd.Metric) {
+		m.Tags = append(m.Tags, t...)
+	}
+}
+
+func DropSource(m *gostatsd.Metric) {
+	m.Source = gostatsd.UnknownSource
+}
+
+func DropTag(t string) MetricOpt {
+	return func(m *gostatsd.Metric) {
+		next := 0
+		found := false
+		for _, tag := range m.Tags {
+			if t == tag {
+				found = true
+				m.Tags[next] = tag
+				next++
+			}
+		}
+		if !found {
+			panic(fmt.Sprintf("failed to find tag %s while building metric", t))
+		}
+		m.Tags = m.Tags[:next]
+	}
+}
+
+// SortCompare func for metrics so they can be compared with require.EqualValues
+// Invoke with sort.Slice(x, SortCompare(x))
+func SortCompare(ms []*gostatsd.Metric) func(i, j int) bool {
+	return func(i, j int) bool {
+		if ms[i].Name == ms[j].Name {
+			if len(ms[i].Tags) == len(ms[j].Tags) { // This is not exactly accurate, but close enough with our data
+				if ms[i].Type == gostatsd.SET {
+					return ms[i].StringValue < ms[j].StringValue
+				} else {
+					return ms[i].Value < ms[j].Value
+				}
+			}
+			return len(ms[i].Tags) < len(ms[j].Tags)
+		}
+		return ms[i].Name < ms[j].Name
+	}
+}

metric_consolidator.go

@@ -7,8 +7,9 @@ import (
 	"github.com/tilinna/clock"
 )
 
-// MetricConsolidator will consolidate metrics randomly in to a slice of MetricMaps, and send the slice to the provided
-// channel.  Run can be started in a long running goroutine to perform flushing, or Flush can be called externally.
+// MetricConsolidator will consolidate metrics randomly in to a slice of MetricMaps, and either send the slice to
+// the provided channel, or make them available synchronously through Drain/Fill. Run can also be started in a long
+// running goroutine to perform flushing, or Flush can be called externally to trigger the channel send.
 //
 // Used to consolidate metrics such as:
 // - counter[name=x, value=1]
@@ -30,9 +31,7 @@ type MetricConsolidator struct {
 func NewMetricConsolidator(spots int, flushInterval time.Duration, sink chan<- []*MetricMap) *MetricConsolidator {
 	mc := &MetricConsolidator{}
 	mc.maps = make(chan *MetricMap, spots)
-	for i := 0; i < spots; i++ {
-		mc.maps <- NewMetricMap()
-	}
+	mc.Fill()
 	mc.flushInterval = flushInterval
 	mc.sink = sink
 	return mc
@@ -52,10 +51,11 @@ func (mc *MetricConsolidator) Run(ctx context.Context) {
 	}
 }
 
-// Flush will collect all the MetricMaps in to a slice, send them to the channel provided, then
-// create new MetricMaps for new metrics to land in.  Not thread-safe.
-func (mc *MetricConsolidator) Flush(ctx context.Context) {
-	var mms []*MetricMap
+// Drain will collect all the MetricMaps in the MetricConsolidator and return them.  If the
+// context.Context is canceled before everything can be collected, they are returned to the
+// MetricConsolidator and nil is returned.
+func (mc *MetricConsolidator) Drain(ctx context.Context) []*MetricMap {
+	mms := make([]*MetricMap, 0, cap(mc.maps))
 	for i := 0; i < cap(mc.maps); i++ {
 		select {
 		case mm := <-mc.maps:
@@ -66,9 +66,19 @@ func (mc *MetricConsolidator) Flush(ctx context.Context) {
 			for _, mm := range mms {
 				mc.maps <- mm
 			}
-			return
+			return nil
 		}
 	}
+	return mms
+}
+
+// Flush will collect all the MetricMaps in to a slice, send them to the channel provided, then
+// create new MetricMaps for new metrics to land in.  Not thread-safe.
+func (mc *MetricConsolidator) Flush(ctx context.Context) {
+	mms := mc.Drain(ctx)
+	if mms == nil {
+		return
+	}
 
 	// Send the collected data to the sink before putting new maps in place.  This allows back-pressure
 	// to propagate through the system, if the sink can't keep up.
@@ -77,6 +87,12 @@ func (mc *MetricConsolidator) Flush(ctx context.Context) {
 	case <-ctx.Done():
 	}
 
+	mc.Fill()
+}
+
+// Fill re-populates the MetricConsolidator with empty MetricMaps, it is the pair to Drain and
+// must be called after a successful Drain, must not be called after a failed Drain.
+func (mc *MetricConsolidator) Fill() {
 	for i := 0; i < cap(mc.maps); i++ {
 		mc.maps <- NewMetricMap()
 	}

metric_map.go

@@ -2,7 +2,6 @@ package gostatsd
 
 import (
 	"bytes"
-	"context"
 	"fmt"
 	"strings"
 
@@ -46,6 +45,14 @@ func (mm *MetricMap) Receive(m *Metric) {
 	m.Done()
 }
 
+func MergeMaps(mms []*MetricMap) *MetricMap {
+	mm := NewMetricMap()
+	for _, mmFrom := range mms {
+		mm.Merge(mmFrom)
+	}
+	return mm
+}
+
 func (mm *MetricMap) Merge(mmFrom *MetricMap) {
 	mmFrom.Counters.Each(func(metricName string, tagsKey string, counterFrom Counter) {
 		v, ok := mm.Counters[metricName]
@@ -359,8 +366,8 @@ func (mm *MetricMap) String() string {
 	return buf.String()
 }
 
-// DispatchMetrics will synthesize Metrics from the MetricMap and push them to the supplied PipelineHandler
-func (mm *MetricMap) DispatchMetrics(ctx context.Context, handler RawMetricHandler) {
+// AsMetrics will synthesize Metrics from the MetricMap and return them as a slice
+func (mm *MetricMap) AsMetrics() []*Metric {
 	var metrics []*Metric
 
 	mm.Counters.Each(func(metricName string, tagsKey string, c Counter) {
@@ -426,5 +433,5 @@ func (mm *MetricMap) DispatchMetrics(ctx context.Context, handler RawMetricHandl
 		}
 	})
 
-	handler.DispatchMetrics(ctx, metrics)
+	return metrics
 }

metric_map_test.go

@@ -143,17 +143,13 @@ func BenchmarkReceives(b *testing.B) {
 }
 
 func TestMetricMapDispatch(t *testing.T) {
-	ctx, done := testContext(t)
-	defer done()
-
 	mm := NewMetricMap()
 	metrics := metricsFixtures()
 	for _, metric := range metrics {
 		mm.Receive(metric)
 	}
-	ch := &capturingHandler{}
 
-	mm.DispatchMetrics(ctx, ch)
+	actual := mm.AsMetrics()
 
 	expected := []*Metric{
 		{Name: "abc.def.g", Value: 3, Rate: 1, Type: GAUGE, Timestamp: 10},
@@ -173,28 +169,27 @@ func TestMetricMapDispatch(t *testing.T) {
 		{Name: "uniq.usr", StringValue: "john", Rate: 1, TagsKey: "baz,foo:bar", Tags: Tags{"baz", "foo:bar"}, Type: SET, Timestamp: 10},
 	}
 
-	cmpSort := func(slice []*Metric) func(i, j int) bool {
-		return func(i, j int) bool {
-			if slice[i].Name == slice[j].Name {
-				if len(slice[i].Tags) == len(slice[j].Tags) { // This is not exactly accurate, but close enough with our data
-					if slice[i].Type == SET {
-						return slice[i].StringValue < slice[j].StringValue
-					} else {
-						return slice[i].Value < slice[j].Value
-					}
+	sort.Slice(actual, SortCompare(actual))
+	sort.Slice(expected, SortCompare(expected))
+
+	require.EqualValues(t, expected, actual)
+}
+
+// Copied from internal/fixtures because dependency loops
+func SortCompare(ms []*Metric) func(i, j int) bool {
+	return func(i, j int) bool {
+		if ms[i].Name == ms[j].Name {
+			if len(ms[i].Tags) == len(ms[j].Tags) { // This is not exactly accurate, but close enough with our data
+				if ms[i].Type == SET {
+					return ms[i].StringValue < ms[j].StringValue
+				} else {
+					return ms[i].Value < ms[j].Value
 				}
-				return len(slice[i].Tags) < len(slice[j].Tags)
 			}
-			return slice[i].Name < slice[j].Name
+			return len(ms[i].Tags) < len(ms[j].Tags)
 		}
+		return ms[i].Name < ms[j].Name
 	}
-
-	actual := ch.GetMetrics()
-
-	sort.Slice(actual, cmpSort(actual))
-	sort.Slice(expected, cmpSort(expected))
-
-	require.EqualValues(t, expected, actual)
 }
 
 func TestMetricMapMerge(t *testing.T) {

metrics.go

@@ -53,6 +53,17 @@ type Metric struct {
 	DoneFunc  func()     // Returns the metric to the pool. May be nil. Call Metric.Done(), not this.
 }
 
+func (m *Metric) AddTagsSetSource(additionalTags Tags, newSource Source) {
+	if len(additionalTags) > 0 {
+		m.Tags = m.Tags.Concat(additionalTags)
+		m.TagsKey = ""
+	}
+	if newSource != m.Source {
+		m.Source = newSource
+		m.TagsKey = ""
+	}
+}
+
 // Reset is used to reset a metric to as clean state, called on re-use from the pool.
 func (m *Metric) Reset() {
 	m.Name = ""