From 02e41f376dafa4a89135a44f39bf37f048b7b230 Mon Sep 17 00:00:00 2001
From: Lei Wang <66336933+wangzlei@users.noreply.github.com>
Date: Tue, 27 Feb 2024 11:09:25 -0800
Subject: [PATCH] Add http2 timouts to close bad TCP connection (#216)

* Add http2 timouts to close bad TCP connection

* Update comment
---
 pkg/conn/conn.go | 22 ++++++++++++++++++----
 1 file changed, 18 insertions(+), 4 deletions(-)

diff --git a/pkg/conn/conn.go b/pkg/conn/conn.go
index b87afca..36add11 100644
--- a/pkg/conn/conn.go
+++ b/pkg/conn/conn.go
@@ -11,13 +11,13 @@ package conn
 
 import (
 	"crypto/tls"
+	"encoding/json"
+	"io/ioutil"
 	"net/http"
 	"net/url"
 	"os"
-	"time"
-	"encoding/json"
-	"io/ioutil"
 	"strings"
+	"time"
 
 	"github.com/aws/aws-sdk-go/aws/awserr"
 	"github.com/aws/aws-sdk-go/aws/credentials"
@@ -63,12 +63,26 @@ func getNewHTTPClient(maxIdle int, requestTimeout int, noVerify bool, proxyAddre
 	transport := &http.Transport{
 		MaxIdleConnsPerHost: maxIdle,
 		TLSClientConfig:     tls,
+		IdleConnTimeout:     90 * time.Second, // Should be longer than PutTelemetryRecords call frequency: 60 seconds
 		Proxy:               http.ProxyURL(proxyURL),
 	}
 
 	// is not enabled by default as we configure TLSClientConfig for supporting SSL to data plane.
 	// http2.ConfigureTransport will setup transport layer to use HTTP2
-	http2.ConfigureTransport(transport)
+	h2transport, err := http2.ConfigureTransports(transport)
+	if err != nil {
+		log.Warnf("Failed to configure HTTP2 transport: %v", err)
+	} else {
+		// Adding timeout settings to the http2 transport to prevent bad tcp connection hanging the requests for too long
+		// See: https://t.corp.amazon.com/P104567981
+		// Doc: https://pkg.go.dev/golang.org/x/net/http2#Transport
+		//  - ReadIdleTimeout is the time before a ping is sent when no frame has been received from a connection
+		//  - PingTimeout is the time before the TCP connection being closed if a Ping response is not received
+		// So in total, if a TCP connection goes bad, it would take the combined time before the TCP connection is closed
+		h2transport.ReadIdleTimeout = 1 * time.Second
+		h2transport.PingTimeout = 2 * time.Second
+	}
+
 	http := &http.Client{
 		Transport: transport,
 		Timeout:   time.Second * time.Duration(requestTimeout),