all: use net.Dial instead of fasthttp.Dial, because fasthttp.Dial limits the number of concurrent dials to 1000

app/vmagent/remotewrite/statconn.go

@@ -5,9 +5,9 @@ import (
 	"net"
 	"strings"
 	"sync/atomic"
+	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/netutil"
-	"github.com/VictoriaMetrics/fasthttp"
 	"github.com/VictoriaMetrics/metrics"
 )
 
@@ -15,11 +15,10 @@ func statDial(network, addr string) (conn net.Conn, err error) {
 	if !strings.HasPrefix(network, "tcp") {
 		return nil, fmt.Errorf("unexpected network passed to statDial: %q; it must start from `tcp`", network)
 	}
-	if netutil.TCP6Enabled() {
+	if !netutil.TCP6Enabled() {
-		conn, err = fasthttp.DialDualStack(addr)
+		network = "tcp4"
-	} else {
-		conn, err = fasthttp.Dial(addr)
 	}
+	conn, err = net.DialTimeout(network, addr, 5*time.Second)
 	dialsTotal.Inc()
 	if err != nil {
 		dialErrors.Inc()

docs/CHANGELOG.md

@@ -2,6 +2,8 @@
 
 # tip
 
+* BUGFIX: vmagent: prevent from `dialing to the given TCP address time out` error when scraping big number of unavailable targets. See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/987
+
 * FEATURE: disable final merge for data for the previous month at the beginning of new month, since it may result in high disk IO and CPU usage. Final merge can be enabled by setting `-finalMergeDelay` command-line flag to positive duration.
 
 

go.mod

@@ -7,7 +7,7 @@ require (
 
 	// Do not use the original github.com/valyala/fasthttp because of issues
 	// like https://github.com/valyala/fasthttp/commit/996610f021ff45fdc98c2ce7884d5fa4e7f9199b
-	github.com/VictoriaMetrics/fasthttp v1.0.10
+	github.com/VictoriaMetrics/fasthttp v1.0.11
 	github.com/VictoriaMetrics/metrics v1.12.3
 	github.com/VictoriaMetrics/metricsql v0.9.1
 	github.com/aws/aws-sdk-go v1.36.23

go.sum

@@ -41,8 +41,8 @@ github.com/BurntSushi/toml v0.3.1/go.mod h1:xHWCNGjB5oqiDr8zfno3MHue2Ht5sIBksp03
 github.com/BurntSushi/xgb v0.0.0-20160522181843-27f122750802/go.mod h1:IVnqGOEym/WlBOVXweHU+Q+/VP0lqqI8lqeDx9IjBqo=
 github.com/VictoriaMetrics/fastcache v1.5.7 h1:4y6y0G8PRzszQUYIQHHssv/jgPHAb5qQuuDNdCbyAgw=
 github.com/VictoriaMetrics/fastcache v1.5.7/go.mod h1:ptDBkNMQI4RtmVo8VS/XwRY6RoTu1dAWCbrk+6WsEM8=
-github.com/VictoriaMetrics/fasthttp v1.0.10 h1:1UbdmWK59j7znylu55r0y66/zTaHbw+Xk+ObSGVywmE=
+github.com/VictoriaMetrics/fasthttp v1.0.11 h1:6XOvE1pF/EhW8qoi7V5qJQJ2rhNV+UGrb1/a9vMbTiw=
-github.com/VictoriaMetrics/fasthttp v1.0.10/go.mod h1:3SeUL4zwB/p/a9aEeRc6gdlbrtNHXBJR6N376EgiSHU=
+github.com/VictoriaMetrics/fasthttp v1.0.11/go.mod h1:3SeUL4zwB/p/a9aEeRc6gdlbrtNHXBJR6N376EgiSHU=
 github.com/VictoriaMetrics/metrics v1.12.2/go.mod h1:Z1tSfPfngDn12bTfZSCqArT3OPY3u88J12hSoOhuiRE=
 github.com/VictoriaMetrics/metrics v1.12.3 h1:Fe6JHC6MSEKa+BtLhPN8WIvS+HKPzMc2evEpNeCGy7I=
 github.com/VictoriaMetrics/metrics v1.12.3/go.mod h1:Z1tSfPfngDn12bTfZSCqArT3OPY3u88J12hSoOhuiRE=

lib/proxy/proxy.go

@@ -7,6 +7,7 @@ import (
 	"fmt"
 	"net"
 	"net/url"
+	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/netutil"
 	"github.com/VictoriaMetrics/fasthttp"
@@ -80,10 +81,12 @@ func (u *URL) NewDialFunc(tlsConfig *tls.Config) (fasthttp.DialFunc, error) {
 }
 
 func defaultDialFunc(addr string) (net.Conn, error) {
+	network := "tcp4"
 	if netutil.TCP6Enabled() {
-		return fasthttp.DialDualStack(addr)
+		network = "tcp"
 	}
-	return fasthttp.Dial(addr)
+	// Do not use fasthttp.Dial because of https://github.com/VictoriaMetrics/VictoriaMetrics/issues/987
+	return net.DialTimeout(network, addr, 5*time.Second)
 }
 
 // sendConnectRequest sends CONNECT request to proxyConn for the given addr and authHeader and returns the established connection to dstAddr.

vendor/github.com/VictoriaMetrics/fasthttp/tcpdialer.go

@@ -190,9 +190,6 @@ func (d *tcpDialer) NewDial(timeout time.Duration) DialFunc {
 			if err == ErrDialTimeout {
 				return nil, err
 			}
-			if err, ok := err.(net.Error); ok && err.Timeout() {
-				return nil, err
-			}
 			idx++
 			n--
 		}
@@ -235,7 +232,7 @@ func tryDial(network string, addr *net.TCPAddr, deadline time.Time, concurrencyC
 	ch := chv.(chan dialResult)
 	go func() {
 		var dr dialResult
-		dr.conn, dr.err = net.DialTimeout(network, addr.String(), timeout)
+		dr.conn, dr.err = net.DialTCP(network, nil, addr)
 		ch <- dr
 		<-concurrencyCh
 	}()

vendor/modules.txt

@@ -10,7 +10,7 @@ cloud.google.com/go/internal/version
 cloud.google.com/go/storage
 # github.com/VictoriaMetrics/fastcache v1.5.7
 github.com/VictoriaMetrics/fastcache
-# github.com/VictoriaMetrics/fasthttp v1.0.10
+# github.com/VictoriaMetrics/fasthttp v1.0.11
 github.com/VictoriaMetrics/fasthttp
 github.com/VictoriaMetrics/fasthttp/fasthttputil
 github.com/VictoriaMetrics/fasthttp/stackless