Use new Prometheus text format parser

cmd/tsdb/Makefile

@@ -1,9 +1,9 @@
 build:
-	@go build .
+	@go1.8rc1 build .
 	
 bench: build 
 	@echo ">> running benchmark"
-	@./tsdb bench write --metrics=$(NUM_METRICS) testdata.100k
+	@./tsdb bench write --metrics=$(NUM_METRICS) testdata.1m
 	@go tool pprof -svg ./tsdb benchout/cpu.prof > benchout/cpuprof.svg
 	@go tool pprof -svg ./tsdb benchout/mem.prof > benchout/memprof.svg
 	@go tool pprof -svg ./tsdb benchout/block.prof > benchout/blockprof.svg

cmd/tsdb/main.go

@@ -9,15 +9,14 @@ import (
 	"path/filepath"
 	"runtime"
 	"runtime/pprof"
-	"sort"
 	"sync"
 	"time"
+	"unsafe"
 
 	"github.com/fabxc/tsdb"
 	"github.com/fabxc/tsdb/labels"
-	dto "github.com/prometheus/client_model/go"
+	promlabels "github.com/prometheus/prometheus/pkg/labels"
-	"github.com/prometheus/common/expfmt"
+	"github.com/prometheus/prometheus/pkg/textparse"
-	"github.com/prometheus/common/model"
 	"github.com/spf13/cobra"
 )
 
@@ -97,7 +96,7 @@ func (b *writeBenchmark) run(cmd *cobra.Command, args []string) {
 	}
 	b.storage = st
 
-	var metrics []model.Metric
+	var metrics []labels.Labels
 
 	measureTime("readData", func() {
 		f, err := os.Open(args[0])
@@ -133,23 +132,11 @@ func (b *writeBenchmark) run(cmd *cobra.Command, args []string) {
 	})
 }
 
-func (b *writeBenchmark) ingestScrapes(metrics []model.Metric, scrapeCount int) error {
+func (b *writeBenchmark) ingestScrapes(lbls []labels.Labels, scrapeCount int) error {
 	var wg sync.WaitGroup
 	var mu sync.Mutex
 	var total uint64
 
-	lbls := make([]labels.Labels, 0, len(metrics))
-
-	for _, m := range metrics {
-		lset := make(labels.Labels, 0, len(m))
-		for k, v := range m {
-			lset = append(lset, labels.Label{Name: string(k), Value: string(v)})
-		}
-		sort.Sort(lset)
-
-		lbls = append(lbls, lset)
-	}
-
 	for i := 0; i < scrapeCount; i += 50 {
 		lbls := lbls
 		for len(lbls) > 0 {
@@ -305,44 +292,30 @@ func measureTime(stage string, f func()) {
 	fmt.Printf(">> completed stage=%s duration=%s\n", stage, time.Since(start))
 }
 
-func readPrometheusLabels(r io.Reader, n int) ([]model.Metric, error) {
+func readPrometheusLabels(r io.Reader, n int) ([]labels.Labels, error) {
-	dec := expfmt.NewDecoder(r, expfmt.FmtProtoText)
+	b, err := ioutil.ReadAll(r)
-
+	if err != nil {
-	var mets []model.Metric
-	fps := map[model.Fingerprint]struct{}{}
-	var mf dto.MetricFamily
-	var dups int
-
-	for i := 0; i < n; {
-		if err := dec.Decode(&mf); err != nil {
-			if err == io.EOF {
-				break
-			}
 		return nil, err
 	}
 
-		for _, m := range mf.GetMetric() {
+	p := textparse.New(b)
-			met := make(model.Metric, len(m.GetLabel())+1)
+	i := 0
-			met["__name__"] = model.LabelValue(mf.GetName())
+	var mets []labels.Labels
+	hashes := map[uint64]struct{}{}
 
-			for _, l := range m.GetLabel() {
+	for p.Next() && i < n {
-				met[model.LabelName(l.GetName())] = model.LabelValue(l.GetValue())
+		m := make(labels.Labels, 0, 10)
-			}
+		p.Metric((*promlabels.Labels)(unsafe.Pointer(&m)))
-			if _, ok := fps[met.Fingerprint()]; ok {
+
-				dups++
+		h := m.Hash()
-			} else {
+		if _, ok := hashes[h]; ok {
-				mets = append(mets, met)
+			continue
-				fps[met.Fingerprint()] = struct{}{}
 		}
+		mets = append(mets, m)
+		hashes[h] = struct{}{}
 		i++
 	}
-	}
+	return mets, p.Err()
-	if dups > 0 {
-		fmt.Println("dropped duplicate metrics:", dups)
-	}
-	fmt.Println("read metrics", len(mets[:n]))
-
-	return mets[:n], nil
 }
 
 func exitWithError(err error) {

head.go

@@ -427,7 +427,6 @@ func (h *headBlock) create(hash uint64, lset labels.Labels) *memSeries {
 		lset: lset,
 		ref:  uint32(len(h.series)),
 	}
-	fmt.Println("create", lset)
 
 	// Allocate empty space until we can insert at the given index.
 	h.series = append(h.series, s)