Golang log.With函数代码示例

func (sl *scrapeLoop) append(samples model.Samples) {
	var (
		numOutOfOrder = 0
		numDuplicates = 0
	)

	for _, s := range samples {
		if err := sl.appender.Append(s); err != nil {
			switch err {
			case local.ErrOutOfOrderSample:
				numOutOfOrder++
				log.With("sample", s).With("error", err).Debug("Sample discarded")
			case local.ErrDuplicateSampleForTimestamp:
				numDuplicates++
				log.With("sample", s).With("error", err).Debug("Sample discarded")
			default:
				log.With("sample", s).With("error", err).Warn("Sample discarded")
			}
		}
	}
	if numOutOfOrder > 0 {
		log.With("numDropped", numOutOfOrder).Warn("Error on ingesting out-of-order samples")
	}
	if numDuplicates > 0 {
		log.With("numDropped", numDuplicates).Warn("Error on ingesting samples with different value but same timestamp")
	}
}

func (sl *scrapeLoop) report(start time.Time, duration time.Duration, err error) {
	sl.scraper.report(start, duration, err)

	ts := model.TimeFromUnixNano(start.UnixNano())

	var health model.SampleValue
	if err == nil {
		health = 1
	}

	healthSample := &model.Sample{
		Metric: model.Metric{
			model.MetricNameLabel: scrapeHealthMetricName,
		},
		Timestamp: ts,
		Value:     health,
	}
	durationSample := &model.Sample{
		Metric: model.Metric{
			model.MetricNameLabel: scrapeDurationMetricName,
		},
		Timestamp: ts,
		Value:     model.SampleValue(float64(duration) / float64(time.Second)),
	}

	if err := sl.reportAppender.Append(healthSample); err != nil {
		log.With("sample", healthSample).With("error", err).Warn("Scrape health sample discarded")
	}
	if err := sl.reportAppender.Append(durationSample); err != nil {
		log.With("sample", durationSample).With("error", err).Warn("Scrape duration sample discarded")
	}
}

// logThrottling handles logging of throttled events and has to be started as a
// goroutine. It stops once s.loopStopping is closed.
//
// Logging strategy: Whenever Throttle() is called and returns true, an signal
// is sent to s.throttled. If that happens for the first time, an Error is
// logged that the storage is now throttled. As long as signals continues to be
// sent via s.throttled at least once per minute, nothing else is logged. Once
// no signal has arrived for a minute, an Info is logged that the storage is not
// throttled anymore. This resets things to the initial state, i.e. once a
// signal arrives again, the Error will be logged again.
func (s *memorySeriesStorage) logThrottling() {
	timer := time.NewTimer(time.Minute)
	timer.Stop()

	// Signal exit of the goroutine. Currently only needed by test code.
	defer close(s.logThrottlingStopped)

	for {
		select {
		case <-s.throttled:
			if !timer.Reset(time.Minute) {
				log.
					With("chunksToPersist", s.getNumChunksToPersist()).
					With("maxChunksToPersist", s.maxChunksToPersist).
					With("memoryChunks", atomic.LoadInt64(&numMemChunks)).
					With("maxToleratedMemChunks", int(float64(s.maxMemoryChunks)*toleranceFactorMemChunks)).
					Error("Storage needs throttling. Scrapes and rule evaluations will be skipped.")
			}
		case <-timer.C:
			log.
				With("chunksToPersist", s.getNumChunksToPersist()).
				With("maxChunksToPersist", s.maxChunksToPersist).
				With("memoryChunks", atomic.LoadInt64(&numMemChunks)).
				With("maxToleratedMemChunks", int(float64(s.maxMemoryChunks)*toleranceFactorMemChunks)).
				Info("Storage does not need throttling anymore.")
		case <-s.loopStopping:
			return
		}
	}
}

// Notify implements the Notifier interface.
func (n *Pushover) Notify(ctx context.Context, as ...*types.Alert) error {
	key, ok := GroupKey(ctx)
	if !ok {
		return fmt.Errorf("group key missing")
	}
	data := n.tmpl.Data(receiver(ctx), groupLabels(ctx), as...)

	log.With("incident", key).Debugln("notifying Pushover")

	var err error
	tmpl := tmplText(n.tmpl, data, &err)

	parameters := url.Values{}
	parameters.Add("token", tmpl(string(n.conf.Token)))
	parameters.Add("user", tmpl(string(n.conf.UserKey)))
	title := tmpl(n.conf.Title)
	message := tmpl(n.conf.Message)
	parameters.Add("title", title)
	if len(title) > 512 {
		title = title[:512]
		log.With("incident", key).Debugf("Truncated title to %q due to Pushover message limit", title)
	}
	if len(title)+len(message) > 512 {
		message = message[:512-len(title)]
		log.With("incident", key).Debugf("Truncated message to %q due to Pushover message limit", message)
	}
	message = strings.TrimSpace(message)
	if message == "" {
		// Pushover rejects empty messages.
		message = "(no details)"
	}
	parameters.Add("message", message)
	parameters.Add("url", tmpl(n.conf.URL))
	parameters.Add("priority", tmpl(n.conf.Priority))
	parameters.Add("retry", fmt.Sprintf("%d", int64(time.Duration(n.conf.Retry).Seconds())))
	parameters.Add("expire", fmt.Sprintf("%d", int64(time.Duration(n.conf.Expire).Seconds())))

	apiURL := "https://api.pushover.net/1/messages.json"
	u, err := url.Parse(apiURL)
	if err != nil {
		return err
	}
	u.RawQuery = parameters.Encode()
	log.With("incident", key).Debugf("Pushover URL = %q", u.String())

	resp, err := ctxhttp.Post(ctx, http.DefaultClient, u.String(), "text/plain", nil)
	if err != nil {
		return err
	}
	defer resp.Body.Close()

	if resp.StatusCode/100 != 2 {
		body, err := ioutil.ReadAll(resp.Body)
		if err != nil {
			return err
		}
		return fmt.Errorf("unexpected status code %v (body: %s)", resp.StatusCode, string(body))
	}
	return nil
}

// calculatePersistenceUrgencyScore calculates and returns an urgency score for
// the speed of persisting chunks. The score is between 0 and 1, where 0 means
// no urgency at all and 1 means highest urgency.
//
// The score is the maximum of the two following sub-scores:
//
// (1) The first sub-score is the number of chunks waiting for persistence
// divided by the maximum number of chunks allowed to be waiting for
// persistence.
//
// (2) If there are more chunks in memory than allowed AND there are more chunks
// waiting for persistence than factorMinChunksToPersist times
// -storage.local.max-chunks-to-persist, then the second sub-score is the
// fraction the number of memory chunks has reached between
// -storage.local.memory-chunks and toleranceFactorForMemChunks times
// -storage.local.memory-chunks.
//
// Should the score ever hit persintenceUrgencyScoreForEnteringRushedMode, the
// storage locks into "rushed mode", in which the returned score is always
// bumped up to 1 until the non-bumped score is below
// persintenceUrgencyScoreForLeavingRushedMode.
//
// This method is not goroutine-safe, but it is only ever called by the single
// goroutine that is in charge of series maintenance. According to the returned
// score, series maintenance should be sped up. If a score of 1 is returned,
// checkpointing based on dirty-series count should be disabled, and series
// files should not by synced anymore provided the user has specified the
// adaptive sync strategy.
func (s *memorySeriesStorage) calculatePersistenceUrgencyScore() float64 {
	s.rushedMtx.Lock()
	defer s.rushedMtx.Unlock()

	var (
		chunksToPersist    = float64(s.getNumChunksToPersist())
		maxChunksToPersist = float64(s.maxChunksToPersist)
		memChunks          = float64(atomic.LoadInt64(&numMemChunks))
		maxMemChunks       = float64(s.maxMemoryChunks)
	)
	score := chunksToPersist / maxChunksToPersist
	if chunksToPersist > maxChunksToPersist*factorMinChunksToPersist {
		score = math.Max(
			score,
			(memChunks/maxMemChunks-1)/(toleranceFactorMemChunks-1),
		)
	}
	if score > 1 {
		score = 1
	}
	s.persistenceUrgencyScore.Set(score)

	if s.rushed {
		// We are already in rushed mode. If the score is still above
		// persintenceUrgencyScoreForLeavingRushedMode, return 1 and
		// leave things as they are.
		if score > persintenceUrgencyScoreForLeavingRushedMode {
			return 1
		}
		// We are out of rushed mode!
		s.rushed = false
		s.rushedMode.Set(0)
		log.
			With("urgencyScore", score).
			With("chunksToPersist", int(chunksToPersist)).
			With("maxChunksToPersist", int(maxChunksToPersist)).
			With("memoryChunks", int(memChunks)).
			With("maxMemoryChunks", int(maxMemChunks)).
			Info("Storage has left rushed mode.")
		return score
	}
	if score > persintenceUrgencyScoreForEnteringRushedMode {
		// Enter rushed mode.
		s.rushed = true
		s.rushedMode.Set(1)
		log.
			With("urgencyScore", score).
			With("chunksToPersist", int(chunksToPersist)).
			With("maxChunksToPersist", int(maxChunksToPersist)).
			With("memoryChunks", int(memChunks)).
			With("maxMemoryChunks", int(maxMemChunks)).
			Warn("Storage has entered rushed mode.")
		return 1
	}
	return score
}

// loadSeriesMapAndHeads loads the fingerprint to memory-series mapping and all
// the chunks contained in the checkpoint (and thus not yet persisted to series
// files). The method is capable of loading the checkpoint format v1 and v2. If
// recoverable corruption is detected, or if the dirty flag was set from the
// beginning, crash recovery is run, which might take a while. If an
// unrecoverable error is encountered, it is returned. Call this method during
// start-up while nothing else is running in storage land. This method is
// utterly goroutine-unsafe.
func (p *persistence) loadSeriesMapAndHeads() (sm *seriesMap, chunksToPersist int64, err error) {
	fingerprintToSeries := make(map[model.Fingerprint]*memorySeries)
	sm = &seriesMap{m: fingerprintToSeries}

	defer func() {
		if p.dirty {
			log.Warn("Persistence layer appears dirty.")
			p.startedDirty.Set(1)
			err = p.recoverFromCrash(fingerprintToSeries)
			if err != nil {
				sm = nil
			}
		} else {
			p.startedDirty.Set(0)
		}
	}()

	hs := newHeadsScanner(p.headsFileName())
	defer hs.close()
	for hs.scan() {
		fingerprintToSeries[hs.fp] = hs.series
	}
	if os.IsNotExist(hs.err) {
		return sm, 0, nil
	}
	if hs.err != nil {
		p.dirty = true
		log.
			With("file", p.headsFileName()).
			With("error", hs.err).
			Error("Error reading heads file.")
		return sm, 0, hs.err
	}
	return sm, hs.chunksToPersistTotal, nil
}

// Notify implements the Notifier interface.
func (n *OpsGenie) Notify(ctx context.Context, as ...*types.Alert) error {
	key, ok := GroupKey(ctx)
	if !ok {
		return fmt.Errorf("group key missing")
	}
	data := n.tmpl.Data(receiver(ctx), groupLabels(ctx), as...)

	log.With("incident", key).Debugln("notifying OpsGenie")

	var err error
	tmpl := tmplText(n.tmpl, data, &err)

	details := make(map[string]string, len(n.conf.Details))
	for k, v := range n.conf.Details {
		details[k] = tmpl(v)
	}

	var (
		msg    interface{}
		apiURL string

		apiMsg = opsGenieMessage{
			APIKey: string(n.conf.APIKey),
			Alias:  key,
		}
		alerts = types.Alerts(as...)
	)
	switch alerts.Status() {
	case model.AlertResolved:
		apiURL = n.conf.APIHost + "v1/json/alert/close"
		msg = &opsGenieCloseMessage{&apiMsg}
	default:
		apiURL = n.conf.APIHost + "v1/json/alert"
		msg = &opsGenieCreateMessage{
			opsGenieMessage: &apiMsg,
			Message:         tmpl(n.conf.Description),
			Details:         details,
			Source:          tmpl(n.conf.Source),
		}
	}
	if err != nil {
		return fmt.Errorf("templating error: %s", err)
	}

	var buf bytes.Buffer
	if err := json.NewEncoder(&buf).Encode(msg); err != nil {
		return err
	}

	resp, err := ctxhttp.Post(ctx, http.DefaultClient, apiURL, contentTypeJSON, &buf)
	if err != nil {
		return err
	}
	resp.Body.Close()

	if resp.StatusCode/100 != 2 {
		return fmt.Errorf("unexpected status code %v", resp.StatusCode)
	}
	return nil
}

// Sync extracts a deduplicated set of Alertmanager endpoints from a list
// of target groups definitions.
func (s *alertmanagerSet) Sync(tgs []*config.TargetGroup) {
	all := []alertmanager{}

	for _, tg := range tgs {
		ams, err := alertmanagerFromGroup(tg, s.cfg)
		if err != nil {
			log.With("err", err).Error("generating discovered Alertmanagers failed")
			continue
		}
		all = append(all, ams...)
	}

	s.mtx.Lock()
	defer s.mtx.Unlock()
	// Set new Alertmanagers and deduplicate them along their unique URL.
	s.ams = []alertmanager{}
	seen := map[string]struct{}{}

	for _, am := range all {
		us := am.url()
		if _, ok := seen[us]; ok {
			continue
		}

		seen[us] = struct{}{}
		s.ams = append(s.ams, am)
	}
}

func init() {
	runtime.ErrorHandlers = []func(error){
		func(err error) {
			log.With("component", "kube_client_runtime").Errorln(err)
		},
	}
}

// NewDispatcher returns a new Dispatcher.
func NewDispatcher(ap provider.Alerts, r *Route, n notify.Notifier, mk types.Marker) *Dispatcher {
	disp := &Dispatcher{
		alerts:   ap,
		notifier: n,
		route:    r,
		marker:   mk,
		log:      log.With("component", "dispatcher"),
	}
	return disp
}

// Notify implements the Notifier interface.
//
// http://developer.pagerduty.com/documentation/integration/events/trigger
func (n *PagerDuty) Notify(ctx context.Context, as ...*types.Alert) error {
	key, ok := GroupKey(ctx)
	if !ok {
		return fmt.Errorf("group key missing")
	}

	var err error
	var (
		alerts    = types.Alerts(as...)
		data      = n.tmpl.Data(receiver(ctx), groupLabels(ctx), as...)
		tmpl      = tmplText(n.tmpl, data, &err)
		eventType = pagerDutyEventTrigger
	)
	if alerts.Status() == model.AlertResolved {
		eventType = pagerDutyEventResolve
	}

	log.With("incident", key).With("eventType", eventType).Debugln("notifying PagerDuty")

	details := make(map[string]string, len(n.conf.Details))
	for k, v := range n.conf.Details {
		details[k] = tmpl(v)
	}

	msg := &pagerDutyMessage{
		ServiceKey:  tmpl(string(n.conf.ServiceKey)),
		EventType:   eventType,
		IncidentKey: key,
		Description: tmpl(n.conf.Description),
		Details:     details,
	}
	if eventType == pagerDutyEventTrigger {
		msg.Client = tmpl(n.conf.Client)
		msg.ClientURL = tmpl(n.conf.ClientURL)
	}
	if err != nil {
		return err
	}

	var buf bytes.Buffer
	if err := json.NewEncoder(&buf).Encode(msg); err != nil {
		return err
	}

	resp, err := ctxhttp.Post(ctx, http.DefaultClient, n.conf.URL, contentTypeJSON, &buf)
	if err != nil {
		return err
	}
	resp.Body.Close()

	if resp.StatusCode/100 != 2 {
		return fmt.Errorf("unexpected status code %v", resp.StatusCode)
	}
	return nil
}

// sendAll sends the alerts to all configured Alertmanagers at concurrently.
// It returns the number of sends that have failed.
func (n *Notifier) sendAll(alerts ...*model.Alert) int {
	begin := time.Now()

	// Attach external labels before sending alerts.
	for _, a := range alerts {
		for ln, lv := range n.opts.ExternalLabels {
			if _, ok := a.Labels[ln]; !ok {
				a.Labels[ln] = lv
			}
		}
	}

	b, err := json.Marshal(alerts)
	if err != nil {
		log.Errorf("Encoding alerts failed: %s", err)
		return len(n.opts.AlertmanagerURLs)
	}
	ctx, _ := context.WithTimeout(context.Background(), n.opts.Timeout)

	send := func(u string) error {
		resp, err := ctxhttp.Post(ctx, http.DefaultClient, postURL(u), contentTypeJSON, bytes.NewReader(b))
		if err != nil {
			return err
		}
		defer resp.Body.Close()

		if resp.StatusCode/100 != 2 {
			return fmt.Errorf("bad response status %v", resp.Status)
		}
		return err
	}

	var (
		wg        sync.WaitGroup
		numErrors uint64
	)
	for _, u := range n.opts.AlertmanagerURLs {
		wg.Add(1)

		go func(u string) {
			if err := send(u); err != nil {
				log.With("alertmanager", u).With("count", fmt.Sprintf("%d", len(alerts))).Errorf("Error sending alerts: %s", err)
				n.errors.WithLabelValues(u).Inc()
				atomic.AddUint64(&numErrors, 1)
			}
			n.latency.WithLabelValues(u).Observe(float64(time.Since(begin)) / float64(time.Second))
			n.sent.WithLabelValues(u).Add(float64(len(alerts)))

			wg.Done()
		}(u)
	}
	wg.Wait()

	return int(numErrors)
}

// setDirty flags the storage as dirty in a goroutine-safe way. The provided
// error will be logged as a reason the first time the storage is flagged as dirty.
func (p *persistence) setDirty(err error) {
	p.dirtyCounter.Inc()
	p.dirtyMtx.Lock()
	defer p.dirtyMtx.Unlock()
	if p.becameDirty {
		return
	}
	p.dirty = true
	p.becameDirty = true
	log.With("error", err).Error("The storage is now inconsistent. Restart Prometheus ASAP to initiate recovery.")
}

func serveStaticAsset(w http.ResponseWriter, req *http.Request) {
	fp := route.Param(route.Context(req), "filepath")
	fp = filepath.Join("web/ui/static", fp)

	info, err := ui.AssetInfo(fp)
	if err != nil {
		log.With("file", fp).Warn("Could not get file info: ", err)
		w.WriteHeader(http.StatusNotFound)
		return
	}
	file, err := ui.Asset(fp)
	if err != nil {
		if err != io.EOF {
			log.With("file", fp).Warn("Could not get file: ", err)
		}
		w.WriteHeader(http.StatusNotFound)
		return
	}

	http.ServeContent(w, req, info.Name(), info.ModTime(), bytes.NewReader(file))
}

func lookupAll(name string, qtype uint16) (*dns.Msg, error) {
	conf, err := dns.ClientConfigFromFile(resolvConf)
	if err != nil {
		return nil, fmt.Errorf("could not load resolv.conf: %s", err)
	}

	client := &dns.Client{}
	response := &dns.Msg{}

	for _, server := range conf.Servers {
		servAddr := net.JoinHostPort(server, conf.Port)
		for _, suffix := range conf.Search {
			response, err = lookup(name, qtype, client, servAddr, suffix, false)
			if err != nil {
				log.
					With("server", server).
					With("name", name).
					With("suffix", suffix).
					With("reason", err).
					Warn("DNS resolution failed.")
				continue
			}
			if len(response.Answer) > 0 {
				return response, nil
			}
		}
		response, err = lookup(name, qtype, client, servAddr, "", false)
		if err == nil {
			return response, nil
		}
		log.
			With("server", server).
			With("name", name).
			With("reason", err).
			Warn("DNS resolution failed.")
	}
	return response, fmt.Errorf("could not resolve %s: no server responded", name)
}