From f082e64e0c36fb929f11bb194c1d9a97cd0d5b48 Mon Sep 17 00:00:00 2001
From: Aliaksandr Valialkin <valyala@victoriametrics.com>
Date: Wed, 6 Apr 2022 12:28:54 +0300
Subject: [PATCH] app/vmagent: reduce the probability of TLS handshake timeout
 when dialing the remote storage

The following actions are taken:

- Increase the TLS hashdshake timeout from 5 seconds to 10 seconds
- Increase dial timeout from 5 seconds to 30 seconds
- Specify DialContext instead of Dial in http.Transport. This allows properly handling
  the Context arg during dialing the remote storage

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1699
---
 app/vmagent/remotewrite/client.go   |  4 ++--
 app/vmagent/remotewrite/statconn.go | 22 ++++++++++++++++++++--
 docs/CHANGELOG.md                   |  1 +
 3 files changed, 23 insertions(+), 4 deletions(-)

diff --git a/app/vmagent/remotewrite/client.go b/app/vmagent/remotewrite/client.go
index 0493758b8..d265ac0ff 100644
--- a/app/vmagent/remotewrite/client.go
+++ b/app/vmagent/remotewrite/client.go
@@ -92,9 +92,9 @@ func newHTTPClient(argIdx int, remoteWriteURL, sanitizedURL string, fq *persiste
 	}
 	tlsCfg := authCfg.NewTLSConfig()
 	tr := &http.Transport{
-		Dial:                statDial,
+		DialContext:         statDial,
 		TLSClientConfig:     tlsCfg,
-		TLSHandshakeTimeout: 5 * time.Second,
+		TLSHandshakeTimeout: 10 * time.Second,
 		MaxConnsPerHost:     2 * concurrency,
 		MaxIdleConnsPerHost: 2 * concurrency,
 		IdleConnTimeout:     time.Minute,
diff --git a/app/vmagent/remotewrite/statconn.go b/app/vmagent/remotewrite/statconn.go
index 5d597e9bf..9aa064c29 100644
--- a/app/vmagent/remotewrite/statconn.go
+++ b/app/vmagent/remotewrite/statconn.go
@@ -1,7 +1,9 @@
 package remotewrite
 
 import (
+	"context"
 	"net"
+	"sync"
 	"sync/atomic"
 	"time"
 
@@ -9,9 +11,25 @@ import (
 	"github.com/VictoriaMetrics/metrics"
 )
 
-func statDial(networkUnused, addr string) (conn net.Conn, err error) {
+func getStdDialer() *net.Dialer {
+	stdDialerOnce.Do(func() {
+		stdDialer = &net.Dialer{
+			Timeout:   30 * time.Second,
+			KeepAlive: 30 * time.Second,
+			DualStack: netutil.TCP6Enabled(),
+		}
+	})
+	return stdDialer
+}
+
+var (
+	stdDialer     *net.Dialer
+	stdDialerOnce sync.Once
+)
+
+func statDial(ctx context.Context, networkUnused, addr string) (conn net.Conn, err error) {
 	network := netutil.GetTCPNetwork()
-	conn, err = net.DialTimeout(network, addr, 5*time.Second)
+	conn, err = stdDialer.DialContext(ctx, network, addr)
 	dialsTotal.Inc()
 	if err != nil {
 		dialErrors.Inc()
diff --git a/docs/CHANGELOG.md b/docs/CHANGELOG.md
index 85a4b010a..cf6cdb785 100644
--- a/docs/CHANGELOG.md
+++ b/docs/CHANGELOG.md
@@ -33,6 +33,7 @@ Previously the `-search.maxUniqueTimeseries` command-line flag was used as a glo
 
 When using [cluster version of VictoriaMetrics](https://docs.victoriametrics.com/Cluster-VictoriaMetrics.html), these command-line flags (including `-search.maxUniqueTimeseries`) must be passed to `vmselect` instead of `vmstorage`.
 
+* BUGFIX: [vmagent](https://docs.victoriametrics.com/vmagent.html) and [vmauth](https://docs.victoriametrics.com/vmauth.html): reduce the probability of `TLS handshake error from XX.XX.XX.XX: EOF` errors when `-remoteWrite.url` points to HTTPS url at `vmauth`. See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1699).
 * BUGFIX: return `Content-Type: text/html` response header when requesting `/` HTTP path at VictoriaMetrics components. Previously `text/plain` response header was returned, which could lead to broken page formatting. See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/2323).
 * BUGFIX: [Graphite Render API](https://docs.victoriametrics.com/#graphite-render-api-usage): accept floating-point values for [maxDataPoints](https://graphite.readthedocs.io/en/stable/render_api.html#maxdatapoints) query arg, since some clients send floating-point values instead of integer values for this arg.