Add per scrape-config targets limit (#7554)

* Add per scrape-config targets limit
Signed-off-by: NJulien Pivotto <roidelapluie@inuits.eu>

config/config.go

@@ -379,8 +379,11 @@ type ScrapeConfig struct {
 	MetricsPath string `yaml:"metrics_path,omitempty"`
 	// The URL scheme with which to fetch metrics from targets.
 	Scheme string `yaml:"scheme,omitempty"`
-	// More than this many samples post metric-relabelling will cause the scrape to fail.
+	// More than this many samples post metric-relabeling will cause the scrape to fail.
 	SampleLimit uint `yaml:"sample_limit,omitempty"`
+	// More than this many targets after the target relabeling will cause the
+	// scrapes to fail.
+	TargetLimit uint `yaml:"target_limit,omitempty"`
 
 	// We cannot do proper Go type embedding below as the parser will then parse
 	// values arbitrarily into the overflow maps of further-down types.

docs/configuration/configuration.md

@@ -252,9 +252,16 @@ metric_relabel_configs:
   [ - <relabel_config> ... ]
 
 # Per-scrape limit on number of scraped samples that will be accepted.
-# If more than this number of samples are present after metric relabelling
+# If more than this number of samples are present after metric relabeling
 # the entire scrape will be treated as failed. 0 means no limit.
 [ sample_limit: <int> | default = 0 ]
+
+# Per-scrape config limit on number of unique targets that will be
+# accepted. If more than this number of targets are present after target
+# relabeling, Prometheus will mark the targets as failed without scraping them.
+# 0 means no limit. This is an experimental feature, this behaviour could
+# change in the future.
+[ target_limit: <int> | default = 0 ]
 ```
 
 Where `<job_name>` must be unique across all scrape configurations.

documentation/prometheus-mixin/alerts.libsonnet

@@ -259,6 +259,20 @@
               description: 'Prometheus %(prometheusName)s has missed {{ printf "%%.0f" $value }} rule group evaluations in the last 5m.' % $._config,
             },
           },
+          {
+            alert: 'PrometheusTargetLimitHit',
+            expr: |||
+              increase(prometheus_target_scrape_pool_exceeded_target_limit_total{%(prometheusSelector)s}[5m]) > 0
+            ||| % $._config,
+            'for': '15m',
+            labels: {
+              severity: 'warning',
+            },
+            annotations: {
+              summary: 'Prometheus has dropped targets because some scrape configs have exceeded the targets limit.',
+              description: 'Prometheus %(prometheusName)s has dropped {{ printf "%%.0f" $value }} targets because the number of targets exceeded the configured target_limit.' % $._config,
+            },
+          },
         ],
       },
     ],

scrape/scrape.go

@@ -81,15 +81,35 @@ var (
 	targetScrapePoolReloads = prometheus.NewCounter(
 		prometheus.CounterOpts{
 			Name: "prometheus_target_scrape_pool_reloads_total",
-			Help: "Total number of scrape loop reloads.",
+			Help: "Total number of scrape pool reloads.",
 		},
 	)
 	targetScrapePoolReloadsFailed = prometheus.NewCounter(
 		prometheus.CounterOpts{
 			Name: "prometheus_target_scrape_pool_reloads_failed_total",
-			Help: "Total number of failed scrape loop reloads.",
+			Help: "Total number of failed scrape pool reloads.",
 		},
 	)
+	targetScrapePoolExceededTargetLimit = prometheus.NewCounter(
+		prometheus.CounterOpts{
+			Name: "prometheus_target_scrape_pool_exceeded_target_limit_total",
+			Help: "Total number of times scrape pools hit the target limit, during sync or config reload.",
+		},
+	)
+	targetScrapePoolTargetLimit = prometheus.NewGaugeVec(
+		prometheus.GaugeOpts{
+			Name: "prometheus_target_scrape_pool_target_limit",
+			Help: "Maximum number of targets allowed in this scrape pool.",
+		},
+		[]string{"scrape_job"},
+	)
+	targetScrapePoolTargetsAdded = prometheus.NewGaugeVec(
+		prometheus.GaugeOpts{
+			Name: "prometheus_target_scrape_pool_targets",
+			Help: "Current number of targets in this scrape pool.",
+		},
+		[]string{"scrape_job"},
+	)
 	targetSyncIntervalLength = prometheus.NewSummaryVec(
 		prometheus.SummaryOpts{
 			Name:       "prometheus_target_sync_length_seconds",
@@ -151,6 +171,9 @@ func init() {
 		targetScrapeSampleDuplicate,
 		targetScrapeSampleOutOfOrder,
 		targetScrapeSampleOutOfBounds,
+		targetScrapePoolExceededTargetLimit,
+		targetScrapePoolTargetLimit,
+		targetScrapePoolTargetsAdded,
 		targetScrapeCacheFlushForced,
 		targetMetadataCache,
 	)
@@ -170,6 +193,7 @@ type scrapePool struct {
 	droppedTargets []*Target
 	loops          map[uint64]loop
 	cancel         context.CancelFunc
+	targetLimitHit bool // Internal state to speed up the target_limit checks.
 
 	// Constructor for new scrape loops. This is settable for testing convenience.
 	newLoop func(scrapeLoopOptions) loop
@@ -278,6 +302,13 @@ func (sp *scrapePool) stop() {
 	}
 	wg.Wait()
 	sp.client.CloseIdleConnections()
+
+	if sp.config != nil {
+		targetScrapePoolSyncsCounter.DeleteLabelValues(sp.config.JobName)
+		targetScrapePoolTargetLimit.DeleteLabelValues(sp.config.JobName)
+		targetScrapePoolTargetsAdded.DeleteLabelValues(sp.config.JobName)
+		targetSyncIntervalLength.DeleteLabelValues(sp.config.JobName)
+	}
 }
 
 // reload the scrape pool with the given scrape configuration. The target state is preserved
@@ -301,6 +332,8 @@ func (sp *scrapePool) reload(cfg *config.ScrapeConfig) error {
 	oldClient := sp.client
 	sp.client = client
 
+	targetScrapePoolTargetLimit.WithLabelValues(sp.config.JobName).Set(float64(sp.config.TargetLimit))
+
 	var (
 		wg              sync.WaitGroup
 		interval        = time.Duration(sp.config.ScrapeInterval)
@@ -311,6 +344,7 @@ func (sp *scrapePool) reload(cfg *config.ScrapeConfig) error {
 		mrc             = sp.config.MetricRelabelConfigs
 	)
 
+	forcedErr := sp.refreshTargetLimitErr()
 	for fp, oldLoop := range sp.loops {
 		var cache *scrapeCache
 		if oc := oldLoop.getCache(); reuseCache && oc != nil {
@@ -338,6 +372,7 @@ func (sp *scrapePool) reload(cfg *config.ScrapeConfig) error {
 			oldLoop.stop()
 			wg.Done()
 
+			newLoop.setForcedError(forcedErr)
 			go newLoop.run(interval, timeout, nil)
 		}(oldLoop, newLoop)
 
@@ -355,10 +390,12 @@ func (sp *scrapePool) reload(cfg *config.ScrapeConfig) error {
 // Sync converts target groups into actual scrape targets and synchronizes
 // the currently running scraper with the resulting set and returns all scraped and dropped targets.
 func (sp *scrapePool) Sync(tgs []*targetgroup.Group) {
+	sp.mtx.Lock()
+	defer sp.mtx.Unlock()
+
 	start := time.Now()
 
 	var all []*Target
-	sp.mtx.Lock()
 	sp.droppedTargets = []*Target{}
 	for _, tg := range tgs {
 		targets, err := targetsFromGroup(tg, sp.config)
@@ -374,7 +411,6 @@ func (sp *scrapePool) Sync(tgs []*targetgroup.Group) {
 			}
 		}
 	}
-	sp.mtx.Unlock()
 	sp.sync(all)
 
 	targetSyncIntervalLength.WithLabelValues(sp.config.JobName).Observe(
@@ -387,11 +423,9 @@ func (sp *scrapePool) Sync(tgs []*targetgroup.Group) {
 // scrape loops for new targets, and stops scrape loops for disappeared targets.
 // It returns after all stopped scrape loops terminated.
 func (sp *scrapePool) sync(targets []*Target) {
-	sp.mtx.Lock()
-	defer sp.mtx.Unlock()
-
+	// This function expects that you have acquired the sp.mtx lock.
 	var (
-		uniqueTargets   = map[uint64]struct{}{}
+		uniqueLoops     = make(map[uint64]loop)
 		interval        = time.Duration(sp.config.ScrapeInterval)
 		timeout         = time.Duration(sp.config.ScrapeTimeout)
 		limit           = int(sp.config.SampleLimit)
@@ -403,7 +437,6 @@ func (sp *scrapePool) sync(targets []*Target) {
 	for _, t := range targets {
 		t := t
 		hash := t.hash()
-		uniqueTargets[hash] = struct{}{}
 
 		if _, ok := sp.activeTargets[hash]; !ok {
 			s := &targetScraper{Target: t, client: sp.client, timeout: timeout}
@@ -419,8 +452,12 @@ func (sp *scrapePool) sync(targets []*Target) {
 			sp.activeTargets[hash] = t
 			sp.loops[hash] = l
 
-			go l.run(interval, timeout, nil)
+			uniqueLoops[hash] = l
 		} else {
+			// This might be a duplicated target.
+			if _, ok := uniqueLoops[hash]; !ok {
+				uniqueLoops[hash] = nil
+			}
 			// Need to keep the most updated labels information
 			// for displaying it in the Service Discovery web page.
 			sp.activeTargets[hash].SetDiscoveredLabels(t.DiscoveredLabels())
@@ -431,7 +468,7 @@ func (sp *scrapePool) sync(targets []*Target) {
 
 	// Stop and remove old targets and scraper loops.
 	for hash := range sp.activeTargets {
-		if _, ok := uniqueTargets[hash]; !ok {
+		if _, ok := uniqueLoops[hash]; !ok {
 			wg.Add(1)
 			go func(l loop) {
 				l.stop()
@@ -443,6 +480,16 @@ func (sp *scrapePool) sync(targets []*Target) {
 		}
 	}
 
+	targetScrapePoolTargetsAdded.WithLabelValues(sp.config.JobName).Set(float64(len(uniqueLoops)))
+	forcedErr := sp.refreshTargetLimitErr()
+	for _, l := range sp.loops {
+		l.setForcedError(forcedErr)
+	}
+	for _, l := range uniqueLoops {
+		if l != nil {
+			go l.run(interval, timeout, nil)
+		}
+	}
 	// Wait for all potentially stopped scrapers to terminate.
 	// This covers the case of flapping targets. If the server is under high load, a new scraper
 	// may be active and tries to insert. The old scraper that didn't terminate yet could still
@@ -450,6 +497,26 @@ func (sp *scrapePool) sync(targets []*Target) {
 	wg.Wait()
 }
 
+// refreshTargetLimitErr returns an error that can be passed to the scrape loops
+// if the number of targets exceeds the configured limit.
+func (sp *scrapePool) refreshTargetLimitErr() error {
+	// This function expects that you have acquired the sp.mtx lock.
+	if sp.config == nil || sp.config.TargetLimit == 0 && !sp.targetLimitHit {
+		return nil
+	}
+	l := len(sp.activeTargets)
+	if l <= int(sp.config.TargetLimit) && !sp.targetLimitHit {
+		return nil
+	}
+	var err error
+	sp.targetLimitHit = l > int(sp.config.TargetLimit)
+	if sp.targetLimitHit {
+		targetScrapePoolExceededTargetLimit.Inc()
+		err = fmt.Errorf("target_limit exceeded (number of targets: %d, limit: %d)", l, sp.config.TargetLimit)
+	}
+	return err
+}
+
 func mutateSampleLabels(lset labels.Labels, target *Target, honor bool, rc []*relabel.Config) labels.Labels {
 	lb := labels.NewBuilder(lset)
 
@@ -590,6 +657,7 @@ func (s *targetScraper) scrape(ctx context.Context, w io.Writer) (string, error)
 // A loop can run and be stopped again. It must not be reused after it was stopped.
 type loop interface {
 	run(interval, timeout time.Duration, errc chan<- error)
+	setForcedError(err error)
 	stop()
 	getCache() *scrapeCache
 	disableEndOfRunStalenessMarkers()
@@ -610,6 +678,8 @@ type scrapeLoop struct {
 	buffers         *pool.Pool
 	jitterSeed      uint64
 	honorTimestamps bool
+	forcedErr       error
+	forcedErrMtx    sync.Mutex
 
 	appender            func() storage.Appender
 	sampleMutator       labelsMutator
@@ -953,10 +1023,7 @@ mainLoop:
 // together with reporting metrics, by using as few appenders as possible.
 // In the happy scenario, a single appender is used.
 func (sl *scrapeLoop) scrapeAndReport(interval, timeout time.Duration, last time.Time, errc chan<- error) time.Time {
-	var (
-		start             = time.Now()
-		scrapeCtx, cancel = context.WithTimeout(sl.ctx, timeout)
-	)
+	start := time.Now()
 
 	// Only record after the first scrape.
 	if !last.IsZero() {
@@ -968,7 +1035,34 @@ func (sl *scrapeLoop) scrapeAndReport(interval, timeout time.Duration, last time
 	b := sl.buffers.Get(sl.lastScrapeSize).([]byte)
 	buf := bytes.NewBuffer(b)
 
-	contentType, scrapeErr := sl.scraper.scrape(scrapeCtx, buf)
+	app := sl.appender()
+	var total, added, seriesAdded int
+	var err, appErr, scrapeErr error
+	defer func() {
+		if err != nil {
+			app.Rollback()
+			return
+		}
+		err = app.Commit()
+		if err != nil {
+			level.Error(sl.l).Log("msg", "Scrape commit failed", "err", err)
+		}
+	}()
+	if forcedErr := sl.getForcedError(); forcedErr != nil {
+		appErr = forcedErr
+		// Add stale markers.
+		if _, _, _, err := sl.append(app, []byte{}, "", start); err != nil {
+			app.Rollback()
+			app = sl.appender()
+			level.Warn(sl.l).Log("msg", "Append failed", "err", err)
+		}
+		if errc != nil {
+			errc <- forcedErr
+		}
+	} else {
+		var contentType string
+		scrapeCtx, cancel := context.WithTimeout(sl.ctx, timeout)
+		contentType, scrapeErr = sl.scraper.scrape(scrapeCtx, buf)
 		cancel()
 
 		if scrapeErr == nil {
@@ -986,21 +1080,9 @@ func (sl *scrapeLoop) scrapeAndReport(interval, timeout time.Duration, last time
 			}
 		}
 
-	app := sl.appender()
-	var err error
-	defer func() {
-		if err != nil {
-			app.Rollback()
-			return
-		}
-		err = app.Commit()
-		if err != nil {
-			level.Error(sl.l).Log("msg", "Scrape commit failed", "err", err)
-		}
-	}()
 		// A failed scrape is the same as an empty scrape,
 		// we still call sl.append to trigger stale markers.
-	total, added, seriesAdded, appErr := sl.append(app, b, contentType, start)
+		total, added, seriesAdded, appErr = sl.append(app, b, contentType, start)
 		if appErr != nil {
 			app.Rollback()
 			app = sl.appender()
@@ -1013,6 +1095,7 @@ func (sl *scrapeLoop) scrapeAndReport(interval, timeout time.Duration, last time
 				level.Warn(sl.l).Log("msg", "Append failed", "err", err)
 			}
 		}
+	}
 
 	sl.buffers.Put(b)
 
@@ -1026,6 +1109,18 @@ func (sl *scrapeLoop) scrapeAndReport(interval, timeout time.Duration, last time
 	return start
 }
 
+func (sl *scrapeLoop) setForcedError(err error) {
+	sl.forcedErrMtx.Lock()
+	defer sl.forcedErrMtx.Unlock()
+	sl.forcedErr = err
+}
+
+func (sl *scrapeLoop) getForcedError() error {
+	sl.forcedErrMtx.Lock()
+	defer sl.forcedErrMtx.Unlock()
+	return sl.forcedErr
+}
+
 func (sl *scrapeLoop) endOfRunStaleness(last time.Time, ticker *time.Ticker, interval time.Duration) {
 	// Scraping has stopped. We want to write stale markers but
 	// the target may be recreated, so we wait just over 2 scrape intervals

scrape/scrape_test.go

@@ -140,6 +140,8 @@ func TestDiscoveredLabelsUpdate(t *testing.T) {
 type testLoop struct {
 	startFunc    func(interval, timeout time.Duration, errc chan<- error)
 	stopFunc     func()
+	forcedErr    error
+	forcedErrMtx sync.Mutex
 }
 
 func (l *testLoop) run(interval, timeout time.Duration, errc chan<- error) {
@@ -149,6 +151,18 @@ func (l *testLoop) run(interval, timeout time.Duration, errc chan<- error) {
 func (l *testLoop) disableEndOfRunStalenessMarkers() {
 }
 
+func (l *testLoop) setForcedError(err error) {
+	l.forcedErrMtx.Lock()
+	defer l.forcedErrMtx.Unlock()
+	l.forcedErr = err
+}
+
+func (l *testLoop) getForcedError() error {
+	l.forcedErrMtx.Lock()
+	defer l.forcedErrMtx.Unlock()
+	return l.forcedErr
+}
+
 func (l *testLoop) stop() {
 	l.stopFunc()
 }
@@ -301,6 +315,92 @@ func TestScrapePoolReload(t *testing.T) {
 	testutil.Equals(t, numTargets, len(sp.loops), "Unexpected number of stopped loops after reload")
 }
 
+func TestScrapePoolTargetLimit(t *testing.T) {
+	// On starting to run, new loops created on reload check whether their preceding
+	// equivalents have been stopped.
+	newLoop := func(opts scrapeLoopOptions) loop {
+		l := &testLoop{
+			startFunc: func(interval, timeout time.Duration, errc chan<- error) {},
+			stopFunc:  func() {},
+		}
+		return l
+	}
+	sp := &scrapePool{
+		appendable:    &nopAppendable{},
+		activeTargets: map[uint64]*Target{},
+		loops:         map[uint64]loop{},
+		newLoop:       newLoop,
+		logger:        nil,
+		client:        http.DefaultClient,
+	}
+
+	var tgs = []*targetgroup.Group{}
+	for i := 0; i < 50; i++ {
+		tgs = append(tgs,
+			&targetgroup.Group{
+				Targets: []model.LabelSet{
+					{model.AddressLabel: model.LabelValue(fmt.Sprintf("127.0.0.1:%d", 9090+i))},
+				},
+			},
+		)
+	}
+
+	var limit uint
+	reloadWithLimit := func(l uint) {
+		limit = l
+		testutil.Ok(t, sp.reload(&config.ScrapeConfig{
+			ScrapeInterval: model.Duration(3 * time.Second),
+			ScrapeTimeout:  model.Duration(2 * time.Second),
+			TargetLimit:    l,
+		}))
+	}
+
+	var targets int
+	loadTargets := func(n int) {
+		targets = n
+		sp.Sync(tgs[:n])
+	}
+
+	validateErrorMessage := func(shouldErr bool) {
+		for _, l := range sp.loops {
+			lerr := l.(*testLoop).getForcedError()
+			if shouldErr {
+				testutil.Assert(t, lerr != nil, "error was expected for %d targets with a limit of %d", targets, limit)
+				testutil.Equals(t, fmt.Sprintf("target_limit exceeded (number of targets: %d, limit: %d)", targets, limit), lerr.Error())
+			} else {
+				testutil.Equals(t, nil, lerr)
+			}
+		}
+	}
+
+	reloadWithLimit(0)
+	loadTargets(50)
+
+	// Simulate an initial config with a limit.
+	sp.config.TargetLimit = 30
+	limit = 30
+	loadTargets(50)
+	validateErrorMessage(true)
+
+	reloadWithLimit(50)
+	validateErrorMessage(false)
+
+	reloadWithLimit(40)
+	validateErrorMessage(true)
+
+	loadTargets(30)
+	validateErrorMessage(false)
+
+	loadTargets(40)
+	validateErrorMessage(false)
+
+	loadTargets(41)
+	validateErrorMessage(true)
+
+	reloadWithLimit(51)
+	validateErrorMessage(false)
+}
+
 func TestScrapePoolAppender(t *testing.T) {
 	cfg := &config.ScrapeConfig{}
 	app := &nopAppendable{}
@@ -595,6 +695,57 @@ func TestScrapeLoopRun(t *testing.T) {
 	}
 }
 
+func TestScrapeLoopForcedErr(t *testing.T) {
+	var (
+		signal = make(chan struct{}, 1)
+		errc   = make(chan error)
+
+		scraper = &testScraper{}
+		app     = func() storage.Appender { return &nopAppender{} }
+	)
+
+	ctx, cancel := context.WithCancel(context.Background())
+	sl := newScrapeLoop(ctx,
+		scraper,
+		nil, nil,
+		nopMutator,
+		nopMutator,
+		app,
+		nil,
+		0,
+		true,
+	)
+
+	forcedErr := fmt.Errorf("forced err")
+	sl.setForcedError(forcedErr)
+
+	scraper.scrapeFunc = func(context.Context, io.Writer) error {
+		t.Fatalf("should not be scraped")
+		return nil
+	}
+
+	go func() {
+		sl.run(time.Second, time.Hour, errc)
+		signal <- struct{}{}
+	}()
+
+	select {
+	case err := <-errc:
+		if err != forcedErr {
+			t.Fatalf("Expected forced error but got: %s", err)
+		}
+	case <-time.After(3 * time.Second):
+		t.Fatalf("Expected forced error but got none")
+	}
+	cancel()
+
+	select {
+	case <-signal:
+	case <-time.After(5 * time.Second):
+		t.Fatalf("Scrape not stopped")
+	}
+}
+
 func TestScrapeLoopMetadata(t *testing.T) {
 	var (
 		signal  = make(chan struct{})