mirror of
https://github.com/golang/go
synced 2024-11-06 06:26:13 -07:00
b378960d5b
This allows us to hide the implementation details of how tags are stored on a context from the normal interface, to allow us to explore more efficient mechanisms. The current storage is not intended as the most efficient choice, this cl is about isolating the API so we can experiment with benchmarks in the future. Change-Id: Ib101416bccd8ecdee269cee636b1564d51e1da8a Reviewed-on: https://go-review.googlesource.com/c/tools/+/222854 Run-TryBot: Ian Cottrell <iancottrell@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Robert Findley <rfindley@google.com>
121 lines
3.3 KiB
Go
121 lines
3.3 KiB
Go
// Copyright 2019 The Go Authors. All rights reserved.
|
|
// Use of this source code is governed by a BSD-style
|
|
// license that can be found in the LICENSE file.
|
|
|
|
package prometheus
|
|
|
|
import (
|
|
"bytes"
|
|
"context"
|
|
"fmt"
|
|
"net/http"
|
|
"sort"
|
|
"sync"
|
|
|
|
"golang.org/x/tools/internal/telemetry/event"
|
|
"golang.org/x/tools/internal/telemetry/metric"
|
|
)
|
|
|
|
func New() *Exporter {
|
|
return &Exporter{}
|
|
}
|
|
|
|
type Exporter struct {
|
|
mu sync.Mutex
|
|
metrics []event.MetricData
|
|
}
|
|
|
|
func (e *Exporter) Metric(ctx context.Context, data event.MetricData) {
|
|
e.mu.Lock()
|
|
defer e.mu.Unlock()
|
|
name := data.Handle()
|
|
// We keep the metrics in name sorted order so the page is stable and easy
|
|
// to read. We do this with an insertion sort rather than sorting the list
|
|
// each time
|
|
index := sort.Search(len(e.metrics), func(i int) bool {
|
|
return e.metrics[i].Handle() >= name
|
|
})
|
|
if index >= len(e.metrics) || e.metrics[index].Handle() != name {
|
|
// we have a new metric, so we need to make a space for it
|
|
old := e.metrics
|
|
e.metrics = make([]event.MetricData, len(old)+1)
|
|
copy(e.metrics, old[:index])
|
|
copy(e.metrics[index+1:], old[index:])
|
|
}
|
|
e.metrics[index] = data
|
|
}
|
|
|
|
func (e *Exporter) header(w http.ResponseWriter, name, description string, isGauge, isHistogram bool) {
|
|
kind := "counter"
|
|
if isGauge {
|
|
kind = "gauge"
|
|
}
|
|
if isHistogram {
|
|
kind = "histogram"
|
|
}
|
|
fmt.Fprintf(w, "# HELP %s %s\n", name, description)
|
|
fmt.Fprintf(w, "# TYPE %s %s\n", name, kind)
|
|
}
|
|
|
|
func (e *Exporter) row(w http.ResponseWriter, name string, group event.TagSet, extra string, value interface{}) {
|
|
fmt.Fprint(w, name)
|
|
buf := &bytes.Buffer{}
|
|
fmt.Fprint(buf, group)
|
|
if extra != "" {
|
|
if buf.Len() > 0 {
|
|
fmt.Fprint(buf, ",")
|
|
}
|
|
fmt.Fprint(buf, extra)
|
|
}
|
|
if buf.Len() > 0 {
|
|
fmt.Fprint(w, "{")
|
|
buf.WriteTo(w)
|
|
fmt.Fprint(w, "}")
|
|
}
|
|
fmt.Fprintf(w, " %v\n", value)
|
|
}
|
|
|
|
func (e *Exporter) Serve(w http.ResponseWriter, r *http.Request) {
|
|
e.mu.Lock()
|
|
defer e.mu.Unlock()
|
|
for _, data := range e.metrics {
|
|
switch data := data.(type) {
|
|
case *metric.Int64Data:
|
|
e.header(w, data.Info.Name, data.Info.Description, data.IsGauge, false)
|
|
for i, group := range data.Groups() {
|
|
e.row(w, data.Info.Name, group, "", data.Rows[i])
|
|
}
|
|
|
|
case *metric.Float64Data:
|
|
e.header(w, data.Info.Name, data.Info.Description, data.IsGauge, false)
|
|
for i, group := range data.Groups() {
|
|
e.row(w, data.Info.Name, group, "", data.Rows[i])
|
|
}
|
|
|
|
case *metric.HistogramInt64Data:
|
|
e.header(w, data.Info.Name, data.Info.Description, false, true)
|
|
for i, group := range data.Groups() {
|
|
row := data.Rows[i]
|
|
for j, b := range data.Info.Buckets {
|
|
e.row(w, data.Info.Name+"_bucket", group, fmt.Sprintf(`le="%v"`, b), row.Values[j])
|
|
}
|
|
e.row(w, data.Info.Name+"_bucket", group, `le="+Inf"`, row.Count)
|
|
e.row(w, data.Info.Name+"_count", group, "", row.Count)
|
|
e.row(w, data.Info.Name+"_sum", group, "", row.Sum)
|
|
}
|
|
|
|
case *metric.HistogramFloat64Data:
|
|
e.header(w, data.Info.Name, data.Info.Description, false, true)
|
|
for i, group := range data.Groups() {
|
|
row := data.Rows[i]
|
|
for j, b := range data.Info.Buckets {
|
|
e.row(w, data.Info.Name+"_bucket", group, fmt.Sprintf(`le="%v"`, b), row.Values[j])
|
|
}
|
|
e.row(w, data.Info.Name+"_bucket", group, `le="+Inf"`, row.Count)
|
|
e.row(w, data.Info.Name+"_count", group, "", row.Count)
|
|
e.row(w, data.Info.Name+"_sum", group, "", row.Sum)
|
|
}
|
|
}
|
|
}
|
|
}
|