SDKOptions timeout configuration not working in dns timeout

[joker-gege]

Describe the bug

When I configured an incorrect DNS(centos7.6 wrong configuration of etc/resolv. conf), I found that Aws::SDKOptions timeout configuration not work。Timeout set 500ms，but when i use HeadObject function will cost 5-7s return

Expected Behavior

The expected behavior is that: the Headobject call can return as timeout set 500ms

Current Behavior

now HeadObject function would take approximately 6-7s

Reproduction Steps

code like this，please use yourself ak，sk，bucketname，endpoint

#include <aws/core/Aws.h>
#include <aws/s3/S3Client.h>
#include <aws/s3/model/HeadObjectRequest.h>
#include <iostream>
#include <aws/core/auth/AWSCredentialsProviderChain.h>
#include <aws/core/client/DefaultRetryStrategy.h>
using namespace Aws;
using namespace Aws::Auth;

int main(int argc, char **argv) {
    Aws::SDKOptions options;
    Aws::InitAPI(options); // Should only be called once.
    int result = 0;
    {
        const Aws::String ak = "xxxxxxxxxxxx";
        const Aws::String sk = "xxxxxxxxxxxx";
        const Aws::String bn = "xxxxxxxxxxxx";
        const Aws::String ep = "xxxxxxxxxxxx";
        const Aws::String objKey = "test/test/test";
        Aws::Client::ClientConfiguration ossOption;

        ossOption.connectTimeoutMs=500;
        ossOption.httpRequestTimeoutMs=500;
        ossOption.requestTimeoutMs=500;
        ossOption.endpointOverride = ep;
        ossOption.scheme = Aws::Http::Scheme::HTTP;
        ossOption.verifySSL = false;
        ossOption.region = Aws::String("bj");
        std::shared_ptr<Aws::Client::RetryStrategy> retry(new Aws::Client::DefaultRetryStrategy(0, 0));
        ossOption.retryStrategy= retry;

        Aws::Auth::AWSCredentials cred(ak, sk);
        auto s3Client_ = new Aws::S3::S3Client(cred, ossOption, Aws::Client::AWSAuthV4Signer::PayloadSigningPolicy::Always, true);
        Aws::S3::Model::HeadObjectRequest request;
        request.SetBucket(bn);
        request.SetKey(objKey);
        Aws::S3::Model::HeadObjectOutcome get_object_outcome = s3Client_->HeadObject(request);
        // Aws::Map<Aws::String, Aws::String> meta;
        if (get_object_outcome.IsSuccess()) {
            std::cout << "success" << std::endl;
        } else {
            std::cout << "failuer: " << get_object_outcome.GetError() << std::endl;
        }
    }
    Aws::ShutdownAPI(options);
    return result;
}

Possible Solution

No response

Additional Information/Context

No response

AWS CPP SDK version used

1.11.34

Compiler and Version used

gcc version 8.3.1

Operating System and version

CentOS 7.6

[joker-gege]

i found this question possible caused by libcurl, the code in CurlHandleContainer.cpp

void CurlHandleContainer::SetDefaultOptionsOnHandle(CURL* handle)
{
    //for timeouts to work in a multi-threaded context,
    //always turn signals off. This also forces dns queries to
    //not be included in the timeout calculations.
    curl_easy_setopt(handle, CURLOPT_NOSIGNAL, 1L);
    curl_easy_setopt(handle, CURLOPT_TIMEOUT_MS, m_httpRequestTimeout);
    curl_easy_setopt(handle, CURLOPT_CONNECTTIMEOUT_MS, m_connectTimeout);
    curl_easy_setopt(handle, CURLOPT_LOW_SPEED_LIMIT, m_lowSpeedLimit);
    curl_easy_setopt(handle, CURLOPT_LOW_SPEED_TIME, m_lowSpeedTime < 1000 ? (m_lowSpeedTime == 0 ? 0 : 1) : m_lowSpeedTime / 1000);
    curl_easy_setopt(handle, CURLOPT_TCP_KEEPALIVE, m_enableTcpKeepAlive ? 1L : 0L);
    curl_easy_setopt(handle, CURLOPT_TCP_KEEPINTVL, m_tcpKeepAliveIntervalMs / 1000);
    curl_easy_setopt(handle, CURLOPT_TCP_KEEPIDLE, m_tcpKeepAliveIntervalMs / 1000);
    curl_easy_setopt(handle, CURLOPT_HTTP_VERSION, ConvertHttpVersion(m_version));
    curl_easy_setopt(handle, CURLOPT_MAXCONNECTS, m_maxPoolSize);
}

[github-actions[bot]]

This issue is now closed. Comments on closed issues are hard for our team to see. If you need more assistance, please open a new issue that references this one.