Optinally print OM created lines (prometheus#1408)

Signed-off-by: Arthur Silva Sens <[email protected]>
Co-authored-by: Arthur Silva Sens <[email protected]>
Signed-off-by: Eugene <[email protected]>

Author: 2 people

examples/createdtimestamps/main.go

@@ -0,0 +1,60 @@
+// Copyright 2022 The Prometheus Authors
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+// A simple example of how to exposed created timestamps in OpenMetrics format.
+
+package main
+
+import (
+	"log"
+	"net/http"
+	"time"
+
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/client_golang/prometheus/promhttp"
+)
+
+func main() {
+	requestDurations := prometheus.NewHistogram(prometheus.HistogramOpts{
+		Name:    "http_request_duration_seconds",
+		Help:    "A histogram of the HTTP request durations in seconds.",
+		Buckets: prometheus.ExponentialBuckets(0.1, 1.5, 5),
+	})
+
+	// Create non-global registry.
+	registry := prometheus.NewRegistry()
+	registry.MustRegister(
+		requestDurations,
+	)
+
+	go func() {
+		for {
+			// Record fictional latency.
+			now := time.Now()
+			requestDurations.Observe(time.Since(now).Seconds())
+			time.Sleep(600 * time.Millisecond)
+		}
+	}()
+
+	// Expose /metrics HTTP endpoint using the created custom registry.
+	http.Handle(
+		"/metrics", promhttp.HandlerFor(
+			registry,
+			promhttp.HandlerOpts{
+				EnableOpenMetrics:                   true,
+				EnableOpenMetricsTextCreatedSamples: true,
+			}),
+	)
+	// To test: curl -H 'Accept: application/openmetrics-text' localhost:8080/metrics
+	log.Fatalln(http.ListenAndServe(":8080", nil))
+}

prometheus/promhttp/http.go

@@ -207,7 +207,13 @@ func HandlerForTransactional(reg prometheus.TransactionalGatherer, opts HandlerO
 		if encodingHeader != string(Identity) {
 			rsp.Header().Set(contentEncodingHeader, encodingHeader)
 		}
-		enc := expfmt.NewEncoder(w, contentType)
+
+		var enc expfmt.Encoder
+		if opts.EnableOpenMetricsTextCreatedSamples {
+			enc = expfmt.NewEncoder(w, contentType, expfmt.WithCreatedLines())
+		} else {
+			enc = expfmt.NewEncoder(w, contentType)
+		}
 
 		// handleError handles the error according to opts.ErrorHandling
 		// and returns true if we have to abort after the handling.
@@ -408,6 +414,21 @@ type HandlerOpts struct {
 	// (which changes the identity of the resulting series on the Prometheus
 	// server).
 	EnableOpenMetrics bool
+	// EnableOpenMetricsTextCreatedSamples specifies if this handler should add, extra, synthetic
+	// Created Timestamps for counters, histograms and summaries, which for the current
+	// version of OpenMetrics are defined as extra series with the same name and "_created"
+	// suffix. See also the OpenMetrics specification for more details
+	// https://github.com/OpenObservability/OpenMetrics/blob/main/specification/OpenMetrics.md#counter-1
+	//
+	// Created timestamps are used to improve the accuracy of reset detection,
+	// but the way it's designed in OpenMetrics 1.0 it also dramatically increases cardinality
+	// if the scraper does not handle those metrics correctly (converting to created timestamp
+	// instead of leaving those series as-is). New OpenMetrics versions might improve
+	// this situation.
+	//
+	// Prometheus introduced the feature flag 'created-timestamp-zero-ingestion'
+	// in version 2.50.0 to handle this situation.
+	EnableOpenMetricsTextCreatedSamples bool
 	// ProcessStartTime allows setting process start timevalue that will be exposed
 	// with "Process-Start-Time-Unix" response header along with the metrics
 	// payload. This allow callers to have efficient transformations to cumulative