support for excluding histograms based on regex, unit conversion and upper range limits

Author: sravotto

README.md

@@ -10,6 +10,7 @@ $ metrics-exporter -config config.yaml
 ```
 The configuration, specified in yaml format specifies the cockroach db URL the proxy connects to and the port the proxy it listens to.
 The log-10 linear format precision is configurable, specifing the lower range (in nanoseconds) and the number of linear bins for each logarithmic bin. 
+Optionally, the user can specify the upper range (in nanoseconds), the unit (seconds,milliseconds,microseconds) to convert the bucket ranges, and a regex expression to exclude matching histograms.
 The tls section allows the user to specify CA, cert and private key to connect to the backend. The same configuration is used to configure the HTTPS endpoint that the proxy listen to.
 
 ```text
@@ -18,6 +19,9 @@ port: 8888
 bucket:
   startns: 100000 
   bins: 10 
+  endns: 100000000000
+  unit: millseconds 
+  exclude: (.*internal)
 tls:
   ca: ./certs/ca.crt
   privatekey: ./certs/client.root.key

internal/lib/config.go

@@ -13,6 +13,7 @@ import (
 	"errors"
 	"io/ioutil"
 	"log"
+	"math"
 	"net/url"
 	"os"
 
@@ -50,11 +51,18 @@ Log10 Bucket Configuration
 
  * Bins: the number of linear buckets for each log10 bucket
  * Startns: The lower range in nanoseconds.
-
+ * Endns: Optional upper range
+ * Exclude: Regex of histogram names to exclude
+ * Unit: Time unit to use for the log10 buckets
+ *
 */
 type BucketConfig struct {
 	Bins    int
 	Startns int
+	// optional
+	Endns   int
+	Exclude string
+	Unit    string
 }
 
 func (b *BucketConfig) checkConfig() error {
@@ -64,6 +72,21 @@ func (b *BucketConfig) checkConfig() error {
 	return errors.New("Invalid Bucket Configuration")
 }
 
+func (b *BucketConfig) UnitDiv() float64 {
+	var div float64 = 1
+	if b.Unit != "" {
+		switch b.Unit {
+		case "seconds":
+			div = math.Pow10(9)
+		case "milliseconds":
+			div = math.Pow10(6)
+		case "microseconds":
+			div = math.Pow10(3)
+		}
+	}
+	return div
+}
+
 /*
 TlsConfig  Configuration
  * Ca: CA certificate file location

internal/lib/histograms.go

@@ -7,28 +7,26 @@
 package lib
 
 import (
-	"bytes"
-	"fmt"
-	"log"
 	"math"
-	"os"
 
 	dto "github.com/prometheus/client_model/go"
-	"github.com/prometheus/common/expfmt"
 	"google.golang.org/protobuf/proto"
 )
 
 type log10Bucket struct {
 	BinNums int
 	Curr    float64
 	Max     float64
+	UnitDiv float64
 }
 
-func createLog10Bucket(start float64, max float64, bins int) *log10Bucket {
+func createLog10Bucket(start float64, max float64, bins int, div float64) *log10Bucket {
 	return &log10Bucket{
 		Curr:    start,
 		Max:     max,
-		BinNums: bins}
+		BinNums: bins,
+		UnitDiv: div,
+	}
 }
 
 // Computes the next bin
@@ -60,8 +58,16 @@ func (currLog10Bucket *log10Bucket) addLog10Buckets(
 	count := currHdrBucket.GetCumulativeCount()
 	// last bucket has le = +Inf.
 	if le == math.Inf(1) {
+		for currLog10Bucket.binUpperBound() < currLog10Bucket.Max {
+			bucket := &dto.Bucket{
+				UpperBound:      proto.Float64(currLog10Bucket.binUpperBound() / currLog10Bucket.UnitDiv),
+				CumulativeCount: proto.Uint64(count),
+			}
+			currLog10Bucket.nextBin()
+			newBuckets = append(newBuckets, bucket)
+		}
 		return append(newBuckets, &dto.Bucket{
-			UpperBound:      proto.Float64(currLog10Bucket.binUpperBound()),
+			UpperBound:      proto.Float64(currLog10Bucket.binUpperBound() / currLog10Bucket.UnitDiv),
 			CumulativeCount: proto.Uint64(count)})
 
 	}
@@ -84,9 +90,11 @@ func (currLog10Bucket *log10Bucket) addLog10Buckets(
 		adj := math.Floor(float64(count-pcount) * (le - currLog10Bucket.binUpperBound()) / (le - ple))
 		res := count - uint64(adj)
 		bucket := &dto.Bucket{
-			UpperBound:      proto.Float64(currLog10Bucket.binUpperBound()),
+			UpperBound:      proto.Float64(currLog10Bucket.binUpperBound() / currLog10Bucket.UnitDiv),
 			CumulativeCount: proto.Uint64(res),
 		}
+		//fmt.Printf("%+v", currLog10Bucket)
+		//fmt.Printf("%+v", bucket)
 		currLog10Bucket.nextBin()
 		newBuckets = append(newBuckets, bucket)
 	}
@@ -102,10 +110,13 @@ func TranslateHistogram(config *BucketConfig, mf *dto.MetricFamily) {
 		max := 0.0
 		if len(m.Histogram.Bucket) >= 2 {
 			max = m.Histogram.Bucket[len(m.Histogram.Bucket)-2].GetUpperBound()
+			if config.Endns > 0 {
+				max = float64(config.Endns)
+			}
 			requiredBuckets = requiredBuckets + int(math.Ceil(math.Log10(float64(max))))*bins
 		}
 		newBuckets := make([]*dto.Bucket, 0, requiredBuckets)
-		currLog10Bucket := createLog10Bucket(float64(config.Startns), max, bins)
+		currLog10Bucket := createLog10Bucket(float64(config.Startns), max, bins, config.UnitDiv())
 
 		for _, curr := range m.GetHistogram().GetBucket() {
 			newBuckets = currLog10Bucket.addLog10Buckets(curr, prev, newBuckets)
@@ -114,21 +125,3 @@ func TranslateHistogram(config *BucketConfig, mf *dto.MetricFamily) {
 		m.Histogram.Bucket = newBuckets
 	}
 }
-
-func TranslateFromFile(config *BucketConfig, filename string) {
-	log.Println("Reading from file :" + filename)
-	var parser expfmt.TextParser
-	var r, err = os.Open(filename)
-	if err != nil {
-		panic("File not found")
-	}
-	metricFamilies, _ := parser.TextToMetricFamilies(r)
-	for _, mf := range metricFamilies {
-		if mf.GetType() == dto.MetricType_HISTOGRAM {
-			TranslateHistogram(config, mf)
-		}
-		var buf bytes.Buffer
-		expfmt.MetricFamilyToText(&buf, mf)
-		fmt.Println(buf.String())
-	}
-}

internal/lib/histograms_test.go

@@ -42,7 +42,7 @@ func TestHistogramConversion(t *testing.T) {
 		var buf bytes.Buffer
 		expfmt.MetricFamilyToText(&buf, mf)
 
-		assert.Equal(buf.String(), output)
+		assert.Equal(output, buf.String())
 	}
 }
 
@@ -58,7 +58,7 @@ func TestMultiStoreConversion(t *testing.T) {
 		TranslateHistogram(config, mf)
 		var buf bytes.Buffer
 		expfmt.MetricFamilyToText(&buf, mf)
-		assert.Equal(buf.String(), multistoreout)
+		assert.Equal(multistoreout, buf.String())
 	}
 
 }
@@ -75,7 +75,7 @@ func TestIdentityConversion(t *testing.T) {
 		TranslateHistogram(config, mf)
 		var buf bytes.Buffer
 		expfmt.MetricFamilyToText(&buf, mf)
-		assert.Equal(buf.String(), output)
+		assert.Equal(output, buf.String())
 	}
 
 }

internal/lib/reader.go

@@ -49,7 +49,7 @@ func (r *MetricsReader) ReadMetrics(ctx context.Context) (map[string]*dto.Metric
 		return nil, err
 	}
 	defer data.Body.Close()
-	if data.StatusCode != 200 {
+	if data.StatusCode != http.StatusOK {
 		return nil, errors.New(data.Status)
 	}
 	var parser expfmt.TextParser

internal/lib/writer.go

@@ -8,33 +8,48 @@ package lib
 
 import (
 	"context"
-	"net/http"
+	"io"
+	"regexp"
 
 	dto "github.com/prometheus/client_model/go"
 	"github.com/prometheus/common/expfmt"
 )
 
 // A MetricsWriter write metrics, after transforming them based on the configuration supplied.
 type MetricsWriter struct {
-	Config *Config
+	Config  *Config
+	Exclude *regexp.Regexp
 }
 
 // Create a MetricsWriter
 func CreateMetricsWriter(config *Config) *MetricsWriter {
+
+	var exp *regexp.Regexp
+	if config.Bucket.Exclude != "" {
+
+		exp = regexp.MustCompile(config.Bucket.Exclude)
+	}
+
 	return &MetricsWriter{
-		Config: config,
+		Config:  config,
+		Exclude: exp,
 	}
 }
 
 // Write the metrics, converting HDR Histogram into Log10 linear histograms.
 func (w *MetricsWriter) WriteMetrics(
 	ctx context.Context,
 	metricFamilies map[string]*dto.MetricFamily,
-	h http.ResponseWriter) {
+	out io.Writer) {
 	for _, mf := range metricFamilies {
 		if mf.GetType() == dto.MetricType_HISTOGRAM {
+			//log.Println("processing " + mf.GetName())
+			if w.Exclude != nil && w.Exclude.MatchString(mf.GetName()) {
+				//	log.Println("Skipping " + mf.GetName())
+				continue
+			}
 			TranslateHistogram(&w.Config.Bucket, mf)
 		}
-		expfmt.MetricFamilyToText(h, mf)
+		expfmt.MetricFamilyToText(out, mf)
 	}
 }

main.go

@@ -16,12 +16,14 @@ import (
 	"fmt"
 	"log"
 	"net/http"
+	"os"
 	"runtime"
 	"runtime/debug"
 
 	"internal/lib"
 
 	"github.com/NYTimes/gziphandler"
+	"github.com/prometheus/common/expfmt"
 )
 
 var (
@@ -44,6 +46,18 @@ func printVersionInfo(buildVersion string) {
 	}
 }
 
+func translateFromFile(ctx context.Context, config *lib.BucketConfig, filename string, writer *lib.MetricsWriter) {
+	log.Println("Reading from file :" + filename)
+	var parser expfmt.TextParser
+	var r, err = os.Open(filename)
+	if err != nil {
+		panic("File not found")
+	}
+	metricFamilies, _ := parser.TextToMetricFamilies(r)
+	writer.WriteMetrics(ctx, metricFamilies, os.Stdout)
+
+}
+
 func main() {
 	configLocation := flag.String("config", "", "YAML configuration")
 	printVersion := flag.Bool("version", false, "print version and exit")
@@ -72,14 +86,17 @@ func main() {
 		}
 	}
 
+	writer := lib.CreateMetricsWriter(config)
+	ctx := context.Background()
+
 	if *localFile != "" {
-		lib.TranslateFromFile(&config.Bucket, *localFile)
+		log.Printf("Reading with:\n%+v\n\n", config)
+		translateFromFile(ctx, &config.Bucket, *localFile, writer)
 		return
 	}
 
 	reader := lib.CreateMetricsReader(config, transport)
-	writer := lib.CreateMetricsWriter(config)
-	ctx := context.Background()
+
 	handler := http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		defer r.Body.Close()