Make the collectorID independent of the desc order.

This is actually the intended behavior, and (as a nice side effect)
makes things cheaper to calculate.

Also, introduce a separator character to avoid hash collisions
(like label values {"ab","c"} vs {"a", "bc"}).

Apply the same principles to signature.go.

Change-Id: I607db544f278ed89684fe5fa11abdbc3e03d3061

model/signature.go

@@ -14,56 +14,84 @@
 package model
 
 import (
+	"bytes"
+	"hash"
 	"hash/fnv"
-	"sort"
 )
 
-// cache the signature of an empty label set.
-var emptyLabelSignature = fnv.New64a().Sum64()
+// SeparatorByte is a byte that cannot occur in valid UTF-8 sequences and is
+// used to separate label names, label values, and other strings from each other
+// when calculating their combined hash value (aka signature aka fingerprint).
+const SeparatorByte byte = 255
 
-// LabelsToSignature provides a way of building a unique signature
-// (i.e., fingerprint) for a given label set sequence.
+var (
+	// cache the signature of an empty label set.
+	emptyLabelSignature = fnv.New64a().Sum64()
+
+	hashAndBufPool = make(chan *hashAndBuf, 1024)
+)
+
+type hashAndBuf struct {
+	h hash.Hash64
+	b bytes.Buffer
+}
+
+func getHashAndBuf() *hashAndBuf {
+	select {
+	case hb := <-hashAndBufPool:
+		return hb
+	default:
+		return &hashAndBuf{h: fnv.New64a()}
+	}
+}
+
+func putHashAndBuf(hb *hashAndBuf) {
+	select {
+	case hashAndBufPool <- hb:
+	default:
+	}
+}
+
+// LabelsToSignature returns a unique signature (i.e., fingerprint) for a given
+// label set.
 func LabelsToSignature(labels map[string]string) uint64 {
 	if len(labels) == 0 {
 		return emptyLabelSignature
 	}
 
-	names := make([]string, 0, len(labels))
-	for name := range labels {
-		names = append(names, name)
+	var result uint64
+	hb := getHashAndBuf()
+	defer putHashAndBuf(hb)
+
+	for k, v := range labels {
+		hb.b.WriteString(k)
+		hb.b.WriteByte(SeparatorByte)
+		hb.b.WriteString(v)
+		hb.h.Write(hb.b.Bytes())
+		result ^= hb.h.Sum64()
+		hb.h.Reset()
+		hb.b.Reset()
 	}
-
-	sort.Strings(names)
-
-	hasher := fnv.New64a()
-
-	for _, name := range names {
-		hasher.Write([]byte(name))
-		hasher.Write([]byte(labels[name]))
-	}
-
-	return hasher.Sum64()
+	return result
 }
 
-// LabelValuesToSignature provides a way of building a unique signature
-// (i.e., fingerprint) for a given set of label's values.
+// LabelValuesToSignature returns a unique signature (i.e., fingerprint) for the
+// values of a given label set.
 func LabelValuesToSignature(labels map[string]string) uint64 {
 	if len(labels) == 0 {
 		return emptyLabelSignature
 	}
 
-	names := make([]string, 0, len(labels))
-	for name := range labels {
-		names = append(names, name)
+	var result uint64
+	hb := getHashAndBuf()
+	defer putHashAndBuf(hb)
+
+	for _, v := range labels {
+		hb.b.WriteString(v)
+		hb.h.Write(hb.b.Bytes())
+		result ^= hb.h.Sum64()
+		hb.h.Reset()
+		hb.b.Reset()
 	}
-
-	sort.Strings(names)
-
-	hasher := fnv.New64a()
-
-	for _, name := range names {
-		hasher.Write([]byte(labels[name]))
-	}
-
-	return hasher.Sum64()
+	return result
 }

model/signature_test.go

@@ -29,7 +29,7 @@ func testLabelsToSignature(t testing.TB) {
 		},
 		{
 			in:  map[string]string{"name": "garland, briggs", "fear": "love is not enough"},
-			out: 12256296522964301276,
+			out: 12952432476264840823,
 		},
 	}
 
@@ -88,11 +88,11 @@ func BenchmarkLabelValuesToSignatureSingle(b *testing.B) {
 }
 
 func BenchmarkLabelValuesToSignatureDouble(b *testing.B) {
-	benchmarkLabelValuesToSignature(b, map[string]string{"first-label": "first-label-value", "second-label": "second-label-value"}, 5670080368112985613)
+	benchmarkLabelValuesToSignature(b, map[string]string{"first-label": "first-label-value", "second-label": "second-label-value"}, 8893559499616767364)
 }
 
 func BenchmarkLabelValuesToSignatureTriple(b *testing.B) {
-	benchmarkLabelValuesToSignature(b, map[string]string{"first-label": "first-label-value", "second-label": "second-label-value", "third-label": "third-label-value"}, 2503588453955211397)
+	benchmarkLabelValuesToSignature(b, map[string]string{"first-label": "first-label-value", "second-label": "second-label-value", "third-label": "third-label-value"}, 1685970066862087833)
 }
 
 func benchmarkLabelToSignature(b *testing.B, l map[string]string, e uint64) {
@@ -108,13 +108,13 @@ func BenchmarkLabelToSignatureScalar(b *testing.B) {
 }
 
 func BenchmarkLabelToSignatureSingle(b *testing.B) {
-	benchmarkLabelToSignature(b, map[string]string{"first-label": "first-label-value"}, 2231159900647003583)
+	benchmarkLabelToSignature(b, map[string]string{"first-label": "first-label-value"}, 5147259542624943964)
 }
 
 func BenchmarkLabelToSignatureDouble(b *testing.B) {
-	benchmarkLabelToSignature(b, map[string]string{"first-label": "first-label-value", "second-label": "second-label-value"}, 14091549261072856487)
+	benchmarkLabelToSignature(b, map[string]string{"first-label": "first-label-value", "second-label": "second-label-value"}, 18269973311206963528)
 }
 
 func BenchmarkLabelToSignatureTriple(b *testing.B) {
-	benchmarkLabelToSignature(b, map[string]string{"first-label": "first-label-value", "second-label": "second-label-value", "third-label": "third-label-value"}, 9120920685107702735)
+	benchmarkLabelToSignature(b, map[string]string{"first-label": "first-label-value", "second-label": "second-label-value", "third-label": "third-label-value"}, 15738406913934009676)
 }

prometheus/desc.go

@@ -134,6 +134,7 @@ func NewDesc(fqName, help string, variableLabels []string, constLabels Labels) *
 	for _, val := range labelValues {
 		b.Reset()
 		b.WriteString(val)
+		b.WriteByte(model.SeparatorByte)
 		h.Write(b.Bytes())
 	}
 	d.id = h.Sum64()
@@ -144,10 +145,12 @@ func NewDesc(fqName, help string, variableLabels []string, constLabels Labels) *
 	h.Reset()
 	b.Reset()
 	b.WriteString(help)
+	b.WriteByte(model.SeparatorByte)
 	h.Write(b.Bytes())
 	for _, labelName := range labelNames {
 		b.Reset()
 		b.WriteString(labelName)
+		b.WriteByte(model.SeparatorByte)
 		h.Write(b.Bytes())
 	}
 	d.dimHash = h.Sum64()

prometheus/registry.go

@@ -21,7 +21,6 @@ package prometheus
 
 import (
 	"bytes"
-	"encoding/binary"
 	"errors"
 	"fmt"
 	"hash/fnv"
@@ -35,6 +34,7 @@ import (
 	"code.google.com/p/goprotobuf/proto"
 
 	"github.com/prometheus/client_golang/_vendor/goautoneg"
+	"github.com/prometheus/client_golang/model"
 	"github.com/prometheus/client_golang/text"
 )
 
@@ -207,8 +207,7 @@ func (r *registry) Register(c Collector) (Collector, error) {
 
 	newDescIDs := map[uint64]struct{}{}
 	newDimHashesByName := map[string]uint64{}
-	collectorIDHash := fnv.New64a()
-	buf := make([]byte, 8)
+	var collectorID uint64 // Just a sum of all desc IDs.
 	var duplicateDescErr error
 
 	r.mtx.Lock()
@@ -227,12 +226,11 @@ func (r *registry) Register(c Collector) (Collector, error) {
 			duplicateDescErr = fmt.Errorf("descriptor %s already exists with the same fully-qualified name and const label values", desc)
 		}
 		// If it is not a duplicate desc in this collector, add it to
-		// the hash.  (We allow duplicate descs within the same
+		// the collectorID.  (We allow duplicate descs within the same
 		// collector, but their existence must be a no-op.)
 		if _, exists := newDescIDs[desc.id]; !exists {
 			newDescIDs[desc.id] = struct{}{}
-			binary.BigEndian.PutUint64(buf, desc.id)
-			collectorIDHash.Write(buf)
+			collectorID += desc.id
 		}
 
 		// Are all the label names and the help string consistent with
@@ -257,7 +255,6 @@ func (r *registry) Register(c Collector) (Collector, error) {
 	if len(newDescIDs) == 0 {
 		return nil, errors.New("collector has no descriptors")
 	}
-	collectorID := collectorIDHash.Sum64()
 	if existing, exists := r.collectorsByID[collectorID]; exists {
 		return existing, errAlreadyReg
 	}
@@ -294,16 +291,13 @@ func (r *registry) Unregister(c Collector) bool {
 	}()
 
 	descIDs := map[uint64]struct{}{}
-	collectorIDHash := fnv.New64a()
-	buf := make([]byte, 8)
+	var collectorID uint64 // Just a sum of the desc IDs.
 	for desc := range descChan {
 		if _, exists := descIDs[desc.id]; !exists {
-			binary.BigEndian.PutUint64(buf, desc.id)
-			collectorIDHash.Write(buf)
+			collectorID += desc.id
 			descIDs[desc.id] = struct{}{}
 		}
 	}
-	collectorID := collectorIDHash.Sum64()
 
 	r.mtx.RLock()
 	if _, exists := r.collectorsByID[collectorID]; !exists {
@@ -488,10 +482,12 @@ func (r *registry) checkConsistency(metricFamily *dto.MetricFamily, dtoMetric *d
 	h := fnv.New64a()
 	var buf bytes.Buffer
 	buf.WriteString(desc.fqName)
+	buf.WriteByte(model.SeparatorByte)
 	h.Write(buf.Bytes())
 	for _, lp := range dtoMetric.Label {
 		buf.Reset()
 		buf.WriteString(lp.GetValue())
+		buf.WriteByte(model.SeparatorByte)
 		h.Write(buf.Bytes())
 	}
 	metricHash := h.Sum64()