Add scrape_body_size_bytes metric (#9569)

Fixes #9520

Signed-off-by: Furkan <furkan.turkal@trendyol.com>

docs/feature_flags.md

@@ -71,6 +71,7 @@ When enabled, for each instance scrape, Prometheus stores a sample in the follow
 - `scrape_timeout_seconds`. The configured `scrape_timeout` for a target. This allows you to measure each target to find out how close they are to timing out with `scrape_duration_seconds / scrape_timeout_seconds`.
 - `scrape_sample_limit`. The configured `sample_limit` for a target. This allows you to measure each target
   to find out how close they are to reaching the limit with `scrape_samples_post_metric_relabeling / scrape_sample_limit`. Note that `scrape_sample_limit` can be zero if there is no limit configured, which means that the query above can return `+Inf` for targets with no limit (as we divide by zero). If you want to query only for targets that do have a sample limit use this query: `scrape_samples_post_metric_relabeling / (scrape_sample_limit > 0)`.
+- `scrape_body_size_bytes`. The uncompressed size of the most recent scrape response, if successful. Scrapes failing because `body_size_limit` is exceeded report `-1`, other scrape failures report `0`.
 
 ## New service discovery manager
 

scrape/scrape.go

@@ -264,7 +264,7 @@ const maxAheadTime = 10 * time.Minute
 
 type labelsMutator func(labels.Labels) labels.Labels
 
-func newScrapePool(cfg *config.ScrapeConfig, app storage.Appendable, jitterSeed uint64, logger log.Logger, reportScrapeTimeout bool) (*scrapePool, error) {
+func newScrapePool(cfg *config.ScrapeConfig, app storage.Appendable, jitterSeed uint64, logger log.Logger, reportExtraMetrics bool) (*scrapePool, error) {
 	targetScrapePools.Inc()
 	if logger == nil {
 		logger = log.NewNopLogger()
@@ -313,7 +313,7 @@ func newScrapePool(cfg *config.ScrapeConfig, app storage.Appendable, jitterSeed
 			opts.labelLimits,
 			opts.interval,
 			opts.timeout,
-			reportScrapeTimeout,
+			reportExtraMetrics,
 		)
 	}
 
@@ -861,7 +861,7 @@ type scrapeLoop struct {
 
 	disabledEndOfRunStalenessMarkers bool
 
-	reportScrapeTimeout bool
+	reportExtraMetrics bool
 }
 
 // scrapeCache tracks mappings of exposed metric strings to label sets and
@@ -1122,7 +1122,7 @@ func newScrapeLoop(ctx context.Context,
 	labelLimits *labelLimits,
 	interval time.Duration,
 	timeout time.Duration,
-	reportScrapeTimeout bool,
+	reportExtraMetrics bool,
 ) *scrapeLoop {
 	if l == nil {
 		l = log.NewNopLogger()
@@ -1149,7 +1149,7 @@ func newScrapeLoop(ctx context.Context,
 		labelLimits:         labelLimits,
 		interval:            interval,
 		timeout:             timeout,
-		reportScrapeTimeout: reportScrapeTimeout,
+		reportExtraMetrics:  reportExtraMetrics,
 	}
 	sl.ctx, sl.cancel = context.WithCancel(ctx)
 
@@ -1238,7 +1238,7 @@ func (sl *scrapeLoop) scrapeAndReport(last, appendTime time.Time, errc chan<- er
 	defer sl.buffers.Put(b)
 	buf := bytes.NewBuffer(b)
 
-	var total, added, seriesAdded int
+	var total, added, seriesAdded, bytes int
 	var err, appErr, scrapeErr error
 
 	app := sl.appender(sl.parentCtx)
@@ -1254,7 +1254,7 @@ func (sl *scrapeLoop) scrapeAndReport(last, appendTime time.Time, errc chan<- er
 	}()
 
 	defer func() {
-		if err = sl.report(app, appendTime, time.Since(start), total, added, seriesAdded, scrapeErr); err != nil {
+		if err = sl.report(app, appendTime, time.Since(start), total, added, seriesAdded, bytes, scrapeErr); err != nil {
 			level.Warn(sl.l).Log("msg", "Appending scrape report failed", "err", err)
 		}
 	}()
@@ -1287,11 +1287,15 @@ func (sl *scrapeLoop) scrapeAndReport(last, appendTime time.Time, errc chan<- er
 		if len(b) > 0 {
 			sl.lastScrapeSize = len(b)
 		}
+		bytes = len(b)
 	} else {
 		level.Debug(sl.l).Log("msg", "Scrape failed", "err", scrapeErr)
 		if errc != nil {
 			errc <- scrapeErr
 		}
+		if errors.Is(scrapeErr, errBodySizeLimit) {
+			bytes = -1
+		}
 	}
 
 	// A failed scrape is the same as an empty scrape,
@@ -1645,9 +1649,10 @@ const (
 	scrapeSeriesAddedMetricName   = "scrape_series_added" + "\xff"
 	scrapeTimeoutMetricName       = "scrape_timeout_seconds" + "\xff"
 	scrapeSampleLimitMetricName   = "scrape_sample_limit" + "\xff"
+	scrapeBodySizeBytesMetricName = "scrape_body_size_bytes" + "\xff"
 )
 
-func (sl *scrapeLoop) report(app storage.Appender, start time.Time, duration time.Duration, scraped, added, seriesAdded int, scrapeErr error) (err error) {
+func (sl *scrapeLoop) report(app storage.Appender, start time.Time, duration time.Duration, scraped, added, seriesAdded, bytes int, scrapeErr error) (err error) {
 	sl.scraper.Report(start, duration, scrapeErr)
 
 	ts := timestamp.FromTime(start)
@@ -1672,13 +1677,16 @@ func (sl *scrapeLoop) report(app storage.Appender, start time.Time, duration tim
 	if err = sl.addReportSample(app, scrapeSeriesAddedMetricName, ts, float64(seriesAdded)); err != nil {
 		return
 	}
-	if sl.reportScrapeTimeout {
+	if sl.reportExtraMetrics {
 		if err = sl.addReportSample(app, scrapeTimeoutMetricName, ts, sl.timeout.Seconds()); err != nil {
 			return
 		}
 		if err = sl.addReportSample(app, scrapeSampleLimitMetricName, ts, float64(sl.sampleLimit)); err != nil {
 			return
 		}
+		if err = sl.addReportSample(app, scrapeBodySizeBytesMetricName, ts, float64(bytes)); err != nil {
+			return
+		}
 	}
 	return
 }
@@ -1703,13 +1711,16 @@ func (sl *scrapeLoop) reportStale(app storage.Appender, start time.Time) (err er
 	if err = sl.addReportSample(app, scrapeSeriesAddedMetricName, ts, stale); err != nil {
 		return
 	}
-	if sl.reportScrapeTimeout {
+	if sl.reportExtraMetrics {
 		if err = sl.addReportSample(app, scrapeTimeoutMetricName, ts, stale); err != nil {
 			return
 		}
 		if err = sl.addReportSample(app, scrapeSampleLimitMetricName, ts, stale); err != nil {
 			return
 		}
+		if err = sl.addReportSample(app, scrapeBodySizeBytesMetricName, ts, stale); err != nil {
+			return
+		}
 	}
 	return
 }