Merge pull request #3698 from krasi-georgiev/scrape-config-reload-fix

Scraper config reload bugfix

cmd/prometheus/main.go

@@ -28,6 +28,7 @@ import (
 	"path/filepath"
 	"runtime"
 	"strings"
+	"sync"
 	"syscall"
 	"time"
 
@@ -286,7 +287,10 @@ func main() {
 	reloaders := []func(cfg *config.Config) error{
 		remoteStorage.ApplyConfig,
 		webHandler.ApplyConfig,
+		// The Scrape and notifier managers need to reload before the Discovery manager as
+		// they need to read the most updated config when receiving the new targets list.
 		notifier.ApplyConfig,
+		scrapeManager.ApplyConfig,
 		func(cfg *config.Config) error {
 			c := make(map[string]sd_config.ServiceDiscoveryConfig)
 			for _, v := range cfg.ScrapeConfigs {
@@ -306,7 +310,6 @@ func main() {
 			}
 			return discoveryManagerNotify.ApplyConfig(c)
 		},
-		scrapeManager.ApplyConfig,
 		func(cfg *config.Config) error {
 			// Get all rule files matching the configuration oaths.
 			var files []string
@@ -328,8 +331,22 @@ func main() {
 	// Start all components while we wait for TSDB to open but only load
 	// initial config and mark ourselves as ready after it completed.
 	dbOpen := make(chan struct{})
-	// Wait until the server is ready to handle reloading
-	reloadReady := make(chan struct{})
+
+	// sync.Once is used to make sure we can close the channel at different execution stages(SIGTERM or when the config is loaded).
+	type closeOnce struct {
+		C     chan struct{}
+		once  sync.Once
+		Close func()
+	}
+	// Wait until the server is ready to handle reloading.
+	reloadReady := &closeOnce{
+		C: make(chan struct{}),
+	}
+	reloadReady.Close = func() {
+		reloadReady.once.Do(func() {
+			close(reloadReady.C)
+		})
+	}
 
 	var g group.Group
 	{
@@ -338,12 +355,16 @@ func main() {
 		cancel := make(chan struct{})
 		g.Add(
 			func() error {
+				// Don't forget to release the reloadReady channel so that waiting blocks can exit normally.
 				select {
 				case <-term:
 					level.Warn(logger).Log("msg", "Received SIGTERM, exiting gracefully...")
+					reloadReady.Close()
+
 				case <-webHandler.Quit():
 					level.Warn(logger).Log("msg", "Received termination request via web service, exiting gracefully...")
 				case <-cancel:
+					reloadReady.Close()
 					break
 				}
 				return nil
@@ -384,6 +405,15 @@ func main() {
 	{
 		g.Add(
 			func() error {
+				// When the scrape manager receives a new targets list
+				// it needs to read a valid config for each job.
+				// It depends on the config being in sync with the discovery manager so
+				// we wait until the config is fully loaded.
+				select {
+				case <-reloadReady.C:
+					break
+				}
+
 				err := scrapeManager.Run(discoveryManagerScrape.SyncCh())
 				level.Info(logger).Log("msg", "Scrape manager stopped")
 				return err
@@ -405,11 +435,8 @@ func main() {
 		g.Add(
 			func() error {
 				select {
-				case <-reloadReady:
+				case <-reloadReady.C:
 					break
-				// In case a shutdown is initiated before the reloadReady is released.
-				case <-cancel:
-					return nil
 				}
 
 				for {
@@ -445,6 +472,7 @@ func main() {
 					break
 				// In case a shutdown is initiated before the dbOpen is released
 				case <-cancel:
+					reloadReady.Close()
 					return nil
 				}
 
@@ -452,9 +480,10 @@ func main() {
 					return fmt.Errorf("Error loading config %s", err)
 				}
 
-				close(reloadReady)
+				reloadReady.Close()
+
 				webHandler.Ready()
-				level.Info(logger).Log("msg", "Server is ready to receive requests.")
+				level.Info(logger).Log("msg", "Server is ready to receive web requests.")
 				<-cancel
 				return nil
 			},
@@ -529,7 +558,16 @@ func main() {
 		// so keep this interrupt after the ruleManager.Stop().
 		g.Add(
 			func() error {
+				// When the notifier manager receives a new targets list
+				// it needs to read a valid config for each job.
+				// It depends on the config being in sync with the discovery manager
+				// so we wait until the config is fully loaded.
+				select {
+				case <-reloadReady.C:
+					break
+				}
 				notifier.Run(discoveryManagerNotify.SyncCh())
+				level.Info(logger).Log("msg", "Notifier manager stopped")
 				return nil
 			},
 			func(err error) {

discovery/manager.go

@@ -16,6 +16,7 @@ package discovery
 import (
 	"context"
 	"fmt"
+	"sync"
 
 	"github.com/go-kit/kit/log"
 	"github.com/go-kit/kit/log/level"
@@ -61,10 +62,10 @@ type poolKey struct {
 func NewManager(logger log.Logger) *Manager {
 	return &Manager{
 		logger:         logger,
-		actionCh:       make(chan func(context.Context)),
 		syncCh:         make(chan map[string][]*targetgroup.Group),
 		targets:        make(map[poolKey]map[string]*targetgroup.Group),
 		discoverCancel: []context.CancelFunc{},
+		ctx:            context.Background(),
 	}
 }
 
@@ -72,7 +73,8 @@ func NewManager(logger log.Logger) *Manager {
 // Targets are grouped by the target set name.
 type Manager struct {
 	logger         log.Logger
-	actionCh       chan func(context.Context)
+	mtx            sync.RWMutex
+	ctx            context.Context
 	discoverCancel []context.CancelFunc
 	// Some Discoverers(eg. k8s) send only the updates for a given target group
 	// so we use map[tg.Source]*targetgroup.Group to know which group to update.
@@ -83,10 +85,9 @@ type Manager struct {
 
 // Run starts the background processing
 func (m *Manager) Run(ctx context.Context) error {
+	m.ctx = ctx
 	for {
 		select {
-		case f := <-m.actionCh:
-			f(ctx)
 		case <-ctx.Done():
 			m.cancelDiscoverers()
 			return ctx.Err()
@@ -101,18 +102,17 @@ func (m *Manager) SyncCh() <-chan map[string][]*targetgroup.Group {
 
 // ApplyConfig removes all running discovery providers and starts new ones using the provided config.
 func (m *Manager) ApplyConfig(cfg map[string]sd_config.ServiceDiscoveryConfig) error {
-	err := make(chan error)
-	m.actionCh <- func(ctx context.Context) {
-		m.cancelDiscoverers()
-		for name, scfg := range cfg {
-			for provName, prov := range m.providersFromConfig(scfg) {
-				m.startProvider(ctx, poolKey{setName: name, provider: provName}, prov)
-			}
+	m.mtx.Lock()
+	defer m.mtx.Unlock()
+
+	m.cancelDiscoverers()
+	for name, scfg := range cfg {
+		for provName, prov := range m.providersFromConfig(scfg) {
+			m.startProvider(m.ctx, poolKey{setName: name, provider: provName}, prov)
 		}
-		close(err)
 	}
 
-	return <-err
+	return nil
 }
 
 func (m *Manager) startProvider(ctx context.Context, poolKey poolKey, worker Discoverer) {
@@ -151,39 +151,32 @@ func (m *Manager) cancelDiscoverers() {
 }
 
 func (m *Manager) updateGroup(poolKey poolKey, tgs []*targetgroup.Group) {
-	done := make(chan struct{})
+	m.mtx.Lock()
+	defer m.mtx.Unlock()
 
-	m.actionCh <- func(ctx context.Context) {
-		for _, tg := range tgs {
-			if tg != nil { // Some Discoverers send nil target group so need to check for it to avoid panics.
-				if _, ok := m.targets[poolKey]; !ok {
-					m.targets[poolKey] = make(map[string]*targetgroup.Group)
-				}
-				m.targets[poolKey][tg.Source] = tg
+	for _, tg := range tgs {
+		if tg != nil { // Some Discoverers send nil target group so need to check for it to avoid panics.
+			if _, ok := m.targets[poolKey]; !ok {
+				m.targets[poolKey] = make(map[string]*targetgroup.Group)
 			}
+			m.targets[poolKey][tg.Source] = tg
 		}
-		close(done)
-
 	}
-	<-done
 }
 
 func (m *Manager) allGroups() map[string][]*targetgroup.Group {
-	tSets := make(chan map[string][]*targetgroup.Group)
-
-	m.actionCh <- func(ctx context.Context) {
-		tSetsAll := map[string][]*targetgroup.Group{}
-		for pkey, tsets := range m.targets {
-			for _, tg := range tsets {
-				// Even if the target group 'tg' is empty we still need to send it to the 'Scrape manager'
-				// to signal that it needs to stop all scrape loops for this target set.
-				tSetsAll[pkey.setName] = append(tSetsAll[pkey.setName], tg)
-			}
+	m.mtx.Lock()
+	defer m.mtx.Unlock()
+
+	tSets := map[string][]*targetgroup.Group{}
+	for pkey, tsets := range m.targets {
+		for _, tg := range tsets {
+			// Even if the target group 'tg' is empty we still need to send it to the 'Scrape manager'
+			// to signal that it needs to stop all scrape loops for this target set.
+			tSets[pkey.setName] = append(tSets[pkey.setName], tg)
 		}
-		tSets <- tSetsAll
 	}
-	return <-tSets
-
+	return tSets
 }
 
 func (m *Manager) providersFromConfig(cfg sd_config.ServiceDiscoveryConfig) map[string]Discoverer {

notifier/notifier.go

@@ -490,7 +490,7 @@ func (n *Notifier) sendOne(ctx context.Context, c *http.Client, url string, b []
 
 // Stop shuts down the notification handler.
 func (n *Notifier) Stop() {
-	level.Info(n.logger).Log("msg", "Stopping notification handler...")
+	level.Info(n.logger).Log("msg", "Stopping notification manager...")
 	n.cancel()
 }
 

retrieval/manager.go

@@ -15,6 +15,8 @@ package retrieval
 
 import (
 	"fmt"
+	"reflect"
+	"sync"
 
 	"github.com/go-kit/kit/log"
 	"github.com/go-kit/kit/log/level"
@@ -35,7 +37,6 @@ func NewScrapeManager(logger log.Logger, app Appendable) *ScrapeManager {
 	return &ScrapeManager{
 		append:        app,
 		logger:        logger,
-		actionCh:      make(chan func()),
 		scrapeConfigs: make(map[string]*config.ScrapeConfig),
 		scrapePools:   make(map[string]*scrapePool),
 		graceShut:     make(chan struct{}),
@@ -49,7 +50,7 @@ type ScrapeManager struct {
 	append        Appendable
 	scrapeConfigs map[string]*config.ScrapeConfig
 	scrapePools   map[string]*scrapePool
-	actionCh      chan func()
+	mtx           sync.RWMutex
 	graceShut     chan struct{}
 }
 
@@ -59,8 +60,6 @@ func (m *ScrapeManager) Run(tsets <-chan map[string][]*targetgroup.Group) error
 
 	for {
 		select {
-		case f := <-m.actionCh:
-			f()
 		case ts := <-tsets:
 			m.reload(ts)
 		case <-m.graceShut:
@@ -79,52 +78,60 @@ func (m *ScrapeManager) Stop() {
 
 // ApplyConfig resets the manager's target providers and job configurations as defined by the new cfg.
 func (m *ScrapeManager) ApplyConfig(cfg *config.Config) error {
-	done := make(chan struct{})
-	m.actionCh <- func() {
-		c := make(map[string]*config.ScrapeConfig)
-		for _, scfg := range cfg.ScrapeConfigs {
-			c[scfg.JobName] = scfg
+	m.mtx.Lock()
+	defer m.mtx.Unlock()
+	c := make(map[string]*config.ScrapeConfig)
+	for _, scfg := range cfg.ScrapeConfigs {
+		c[scfg.JobName] = scfg
+	}
+	m.scrapeConfigs = c
+
+	// Cleanup and reload pool if config has changed.
+	for name, sp := range m.scrapePools {
+		if cfg, ok := m.scrapeConfigs[name]; !ok {
+			sp.stop()
+			delete(m.scrapePools, name)
+		} else if !reflect.DeepEqual(sp.config, cfg) {
+			sp.reload(cfg)
 		}
-		m.scrapeConfigs = c
-		close(done)
 	}
-	<-done
+
 	return nil
 }
 
 // TargetMap returns map of active and dropped targets and their corresponding scrape config job name.
 func (m *ScrapeManager) TargetMap() map[string][]*Target {
-	targetsMap := make(chan map[string][]*Target)
-	m.actionCh <- func() {
-		targets := make(map[string][]*Target)
-		for jobName, sp := range m.scrapePools {
-			sp.mtx.RLock()
-			for _, t := range sp.targets {
-				targets[jobName] = append(targets[jobName], t)
-			}
-			targets[jobName] = append(targets[jobName], sp.droppedTargets...)
-			sp.mtx.RUnlock()
+	m.mtx.Lock()
+	defer m.mtx.Unlock()
+
+	targets := make(map[string][]*Target)
+	for jobName, sp := range m.scrapePools {
+		sp.mtx.RLock()
+		for _, t := range sp.targets {
+			targets[jobName] = append(targets[jobName], t)
 		}
-		targetsMap <- targets
+		targets[jobName] = append(targets[jobName], sp.droppedTargets...)
+		sp.mtx.RUnlock()
 	}
-	return <-targetsMap
+
+	return targets
 }
 
 // Targets returns the targets currently being scraped.
 func (m *ScrapeManager) Targets() []*Target {
-	targets := make(chan []*Target)
-	m.actionCh <- func() {
-		var t []*Target
-		for _, p := range m.scrapePools {
-			p.mtx.RLock()
-			for _, tt := range p.targets {
-				t = append(t, tt)
-			}
-			p.mtx.RUnlock()
+	m.mtx.Lock()
+	defer m.mtx.Unlock()
+
+	var targets []*Target
+	for _, p := range m.scrapePools {
+		p.mtx.RLock()
+		for _, tt := range p.targets {
+			targets = append(targets, tt)
 		}
-		targets <- t
+		p.mtx.RUnlock()
 	}
-	return <-targets
+
+	return targets
 }
 
 func (m *ScrapeManager) reload(t map[string][]*targetgroup.Group) {
@@ -146,12 +153,4 @@ func (m *ScrapeManager) reload(t map[string][]*targetgroup.Group) {
 			existing.Sync(tgroup)
 		}
 	}
-
-	// Cleanup - check the config and cancel the scrape loops if it don't exist in the scrape config.
-	for name, sp := range m.scrapePools {
-		if _, ok := m.scrapeConfigs[name]; !ok {
-			sp.stop()
-			delete(m.scrapePools, name)
-		}
-	}
 }