Add proxy_url parameter to allow specifying per-job HTTP proxy servers

Allow scrape_configs to have an optional proxy_url option which specifies
a proxy to be used for all connections to hosts in that config.

Internally this modifies the various client functions to take a *url.URL pointer
which currently must point to an HTTP proxy (but has been left open-ended to
allow the url format to be extended to support others, such as maybe SOCKS if
needed).

config/config.go

@@ -100,6 +100,34 @@ var (
 	}
 )
 
+// This custom URL type allows validating at configuration load time.
+type URL struct {
+	*url.URL
+}
+
+// UnmarshalYAML implements the yaml.Unmarshaler interface for URLs.
+func (u *URL) UnmarshalYAML(unmarshal func(interface{}) error) error {
+	var s string
+	if err := unmarshal(&s); err != nil {
+		return err
+	}
+
+	urlp, err := url.Parse(s)
+	if err != nil {
+		return err
+	}
+	u.URL = urlp
+	return nil
+}
+
+// MarshalYAML implements the yaml.Marshaler interface for URLs.
+func (u URL) MarshalYAML() (interface{}, error) {
+	if u.URL != nil {
+		return u.String(), nil
+	}
+	return nil, nil
+}
+
 // Config is the top-level configuration for Prometheus's config files.
 type Config struct {
 	GlobalConfig  GlobalConfig    `yaml:"global"`
@@ -237,6 +265,8 @@ type ScrapeConfig struct {
 	CACert string `yaml:"ca_cert,omitempty"`
 	// The client cert authentication credentials for the targets.
 	ClientCert *ClientCert `yaml:"client_cert,omitempty"`
+	// HTTP proxy server to use to connect to the targets.
+	ProxyURL URL `yaml:"proxy_url,omitempty"`
 
 	// List of labeled target groups for this job.
 	TargetGroups []*TargetGroup `yaml:"target_groups,omitempty"`

notification/notification.go

@@ -103,7 +103,7 @@ func NewNotificationHandler(o *NotificationHandlerOptions) *NotificationHandler
 		alertmanagerURL:      strings.TrimRight(o.AlertmanagerURL, "/"),
 		pendingNotifications: make(chan NotificationReqs, o.QueueCapacity),
 
-		httpClient: httputil.NewDeadlineClient(o.Deadline),
+		httpClient: httputil.NewDeadlineClient(o.Deadline, nil),
 
 		notificationLatency: prometheus.NewSummary(prometheus.SummaryOpts{
 			Namespace: namespace,

retrieval/target.go

@@ -272,7 +272,7 @@ func newHTTPClient(cfg *config.ScrapeConfig) (*http.Client, error) {
 	tlsConfig.BuildNameToCertificate()
 
 	// Get a default roundtripper with the scrape timeout.
-	rt := httputil.NewDeadlineRoundTripper(time.Duration(cfg.ScrapeTimeout))
+	rt := httputil.NewDeadlineRoundTripper(time.Duration(cfg.ScrapeTimeout), cfg.ProxyURL.URL)
 	tr := rt.(*http.Transport)
 	// Set the TLS config from above
 	tr.TLSClientConfig = tlsConfig

retrieval/target_test.go

@@ -447,7 +447,7 @@ func newTestTarget(targetURL string, deadline time.Duration, baseLabels clientmo
 		deadline:        deadline,
 		status:          &TargetStatus{},
 		scrapeInterval:  1 * time.Millisecond,
-		httpClient:      httputil.NewDeadlineClient(deadline),
+		httpClient:      httputil.NewDeadlineClient(deadline, nil),
 		scraperStopping: make(chan struct{}),
 		scraperStopped:  make(chan struct{}),
 	}

storage/remote/influxdb/client.go

@@ -47,7 +47,7 @@ type Client struct {
 func NewClient(url string, timeout time.Duration, database, retentionPolicy string) *Client {
 	return &Client{
 		url:             url,
-		httpClient:      httputil.NewDeadlineClient(timeout),
+		httpClient:      httputil.NewDeadlineClient(timeout, nil),
 		retentionPolicy: retentionPolicy,
 		database:        database,
 	}

storage/remote/opentsdb/client.go

@@ -50,7 +50,7 @@ type Client struct {
 func NewClient(url string, timeout time.Duration) *Client {
 	return &Client{
 		url:        url,
-		httpClient: httputil.NewDeadlineClient(timeout),
+		httpClient: httputil.NewDeadlineClient(timeout, nil),
 	}
 }
 

util/httputil/client.go

@@ -16,6 +16,7 @@ package httputil
 import (
 	"net"
 	"net/http"
+	"net/url"
 	"time"
 )
 
@@ -26,14 +27,16 @@ func NewClient(rt http.RoundTripper) *http.Client {
 
 // NewDeadlineClient returns a new http.Client which will time out long running
 // requests.
-func NewDeadlineClient(timeout time.Duration) *http.Client {
-	return NewClient(NewDeadlineRoundTripper(timeout))
+func NewDeadlineClient(timeout time.Duration, proxyURL *url.URL) *http.Client {
+	return NewClient(NewDeadlineRoundTripper(timeout, proxyURL))
 }
 
 // NewDeadlineRoundTripper returns a new http.RoundTripper which will time out
 // long running requests.
-func NewDeadlineRoundTripper(timeout time.Duration) http.RoundTripper {
+func NewDeadlineRoundTripper(timeout time.Duration, proxyURL *url.URL) http.RoundTripper {
 	return &http.Transport{
+		// Set proxy (if null, then becomes a direct connection)
+		Proxy: http.ProxyURL(proxyURL),
 		// We need to disable keepalive, because we set a deadline on the
 		// underlying connection.
 		DisableKeepAlives: true,