Add support for k8s service DNS discovery to prometheus input (#3236)

plugins/inputs/prometheus/prometheus.go

@@ -4,7 +4,10 @@ import (
 	"errors"
 	"fmt"
 	"io/ioutil"
+	"log"
+	"net"
 	"net/http"
+	"net/url"
 	"sync"
 	"time"
 
@@ -16,8 +19,12 @@ import (
 const acceptHeader = `application/vnd.google.protobuf;proto=io.prometheus.client.MetricFamily;encoding=delimited;q=0.7,text/plain;version=0.0.4;q=0.3`
 
 type Prometheus struct {
+	// An array of urls to scrape metrics from.
 	Urls []string
 
+	// An array of Kubernetes services to scrape metrics from.
+	KubernetesServices []string
+
 	// Bearer Token authorization file path
 	BearerToken string `toml:"bearer_token"`
 
@@ -39,6 +46,9 @@ var sampleConfig = `
   ## An array of urls to scrape metrics from.
   urls = ["http://localhost:9100/metrics"]
 
+  ## An array of Kubernetes services to scrape metrics from.
+  kubernetes_services = ["http://my-service-dns.my-namespace:9100/metrics"]
+
   ## Use bearer token for authorization
   # bearer_token = /path/to/bearer/token
 
@@ -63,6 +73,53 @@ func (p *Prometheus) Description() string {
 
 var ErrProtocolError = errors.New("prometheus protocol error")
 
+func (p *Prometheus) AddressToURL(u *url.URL, address string) string {
+	host := address
+	if u.Port() != "" {
+		host = address + ":" + u.Port()
+	}
+	reconstructedUrl := url.URL{
+		Scheme:     u.Scheme,
+		Opaque:     u.Opaque,
+		User:       u.User,
+		Path:       u.Path,
+		RawPath:    u.RawPath,
+		ForceQuery: u.ForceQuery,
+		RawQuery:   u.RawQuery,
+		Fragment:   u.Fragment,
+		Host:       host,
+	}
+	return reconstructedUrl.String()
+}
+
+type UrlAndAddress struct {
+	Url     string
+	Address string
+}
+
+func (p *Prometheus) GetAllURLs() ([]UrlAndAddress, error) {
+	allUrls := make([]UrlAndAddress, 0)
+	for _, url := range p.Urls {
+		allUrls = append(allUrls, UrlAndAddress{Url: url})
+	}
+	for _, service := range p.KubernetesServices {
+		u, err := url.Parse(service)
+		if err != nil {
+			return nil, err
+		}
+		resolvedAddresses, err := net.LookupHost(u.Hostname())
+		if err != nil {
+			log.Printf("prometheus: Could not resolve %s, skipping it. Error: %s", u.Host, err)
+			continue
+		}
+		for _, resolved := range resolvedAddresses {
+			serviceUrl := p.AddressToURL(u, resolved)
+			allUrls = append(allUrls, UrlAndAddress{Url: serviceUrl, Address: resolved})
+		}
+	}
+	return allUrls, nil
+}
+
 // Reads stats from all configured servers accumulates stats.
 // Returns one of the errors encountered while gather stats (if any).
 func (p *Prometheus) Gather(acc telegraf.Accumulator) error {
@@ -76,12 +133,16 @@ func (p *Prometheus) Gather(acc telegraf.Accumulator) error {
 
 	var wg sync.WaitGroup
 
-	for _, serv := range p.Urls {
+	allUrls, err := p.GetAllURLs()
+	if err != nil {
+		return err
+	}
+	for _, url := range allUrls {
 		wg.Add(1)
-		go func(serv string) {
+		go func(serviceUrl UrlAndAddress) {
 			defer wg.Done()
-			acc.AddError(p.gatherURL(serv, acc))
-		}(serv)
+			acc.AddError(p.gatherURL(serviceUrl, acc))
+		}(url)
 	}
 
 	wg.Wait()
@@ -116,8 +177,8 @@ func (p *Prometheus) createHttpClient() (*http.Client, error) {
 	return client, nil
 }
 
-func (p *Prometheus) gatherURL(url string, acc telegraf.Accumulator) error {
-	var req, err = http.NewRequest("GET", url, nil)
+func (p *Prometheus) gatherURL(url UrlAndAddress, acc telegraf.Accumulator) error {
+	var req, err = http.NewRequest("GET", url.Url, nil)
 	req.Header.Add("Accept", acceptHeader)
 	var token []byte
 	var resp *http.Response
@@ -132,11 +193,11 @@ func (p *Prometheus) gatherURL(url string, acc telegraf.Accumulator) error {
 
 	resp, err = p.client.Do(req)
 	if err != nil {
-		return fmt.Errorf("error making HTTP request to %s: %s", url, err)
+		return fmt.Errorf("error making HTTP request to %s: %s", url.Url, err)
 	}
 	defer resp.Body.Close()
 	if resp.StatusCode != http.StatusOK {
-		return fmt.Errorf("%s returned HTTP status %s", url, resp.Status)
+		return fmt.Errorf("%s returned HTTP status %s", url.Url, resp.Status)
 	}
 
 	body, err := ioutil.ReadAll(resp.Body)
@@ -147,12 +208,15 @@ func (p *Prometheus) gatherURL(url string, acc telegraf.Accumulator) error {
 	metrics, err := Parse(body, resp.Header)
 	if err != nil {
 		return fmt.Errorf("error reading metrics for %s: %s",
-			url, err)
+			url.Url, err)
 	}
 	// Add (or not) collected metrics
 	for _, metric := range metrics {
 		tags := metric.Tags()
-		tags["url"] = url
+		tags["url"] = url.Url
+		if url.Address != "" {
+			tags["address"] = url.Address
+		}
 		acc.AddFields(metric.Name(), metric.Fields(), tags, metric.Time())
 	}