Apache HttpClient CoreConnectionPNames.CONNECTION_TIMEOUT 什么都不做?
我从 HttpClient 得到奇怪的行为,参数 CoreConnectionPNames.CONNECTION_TIMEOUT 设置为 1。 我预计 HttpGet 请求会失败,抛出连接超时异常,但它们是成功的。这看起来不合理,因为它实际上意味着 TCP 握手在不到 1 毫秒的时间内完成。
我正在使用的 httpclient 版本可以在这个 pom.xml 中看到,
<dependency>
<groupId>org.apache.httpcomponents</groupId>
<artifactId>httpclient</artifactId>
<version>4.0.1</version>
<type>jar</type>
</dependency>
这是代码:
import java.io.IOException;
import java.util.Random;
import org.apache.http.HttpEntity;
import org.apache.http.HttpResponse;
import org.apache.http.client.ClientProtocolException;
import org.apache.http.client.HttpClient;
import org.apache.http.client.methods.HttpGet;
import org.apache.http.client.methods.HttpUriRequest;
import org.apache.http.impl.client.DefaultHttpClient;
import org.apache.http.params.CoreConnectionPNames;
import org.apache.http.util.EntityUtils;
import org.apache.log4j.Logger;
public class TestNodeAliveness {
private static Logger log = Logger.getLogger(TestNodeAliveness.class);
public static boolean nodeBIT(String elasticIP) throws ClientProtocolException, IOException {
try {
HttpClient client = new DefaultHttpClient();
// The time it takes to open TCP connection.
client.getParams().setParameter(CoreConnectionPNames.CONNECTION_TIMEOUT, 1);
// Timeout when server does not send data.
client.getParams().setParameter(CoreConnectionPNames.SO_TIMEOUT, 5000);
// Some tuning that is not required for bit tests.
client.getParams().setParameter(CoreConnectionPNames.STALE_CONNECTION_CHECK, false);
client.getParams().setParameter(CoreConnectionPNames.TCP_NODELAY, true);
HttpUriRequest request = new HttpGet("http://" + elasticIP);
HttpResponse response = client.execute(request);
HttpEntity entity = response.getEntity();
if(entity == null) {
return false;
} else {
System.out.println(EntityUtils.toString(entity));
}
// Close just in case.
request.abort();
} catch (Throwable e) {
log.warn("BIT Test failed for " + elasticIP);
e.printStackTrace();
return false;
}
return true;
}
public static void main(String[] args) throws ClientProtocolException, IOException {
nodeBIT("google.com?cant_cache_this=" + (new Random()).nextInt());
}
}
这怎么可能? 谢谢。
I get strange behavior from HttpClient with parameter CoreConnectionPNames.CONNECTION_TIMEOUT set to 1.
I would expect that HttpGet request would fail, throwing connection timeout exception and yet they are successful. This seems irrational as it actually means that TCP handshake finished in less then 1 millisecond.
The httpclient version I'm using as can be seen in this pom.xml is
<dependency>
<groupId>org.apache.httpcomponents</groupId>
<artifactId>httpclient</artifactId>
<version>4.0.1</version>
<type>jar</type>
</dependency>
Here is the code:
import java.io.IOException;
import java.util.Random;
import org.apache.http.HttpEntity;
import org.apache.http.HttpResponse;
import org.apache.http.client.ClientProtocolException;
import org.apache.http.client.HttpClient;
import org.apache.http.client.methods.HttpGet;
import org.apache.http.client.methods.HttpUriRequest;
import org.apache.http.impl.client.DefaultHttpClient;
import org.apache.http.params.CoreConnectionPNames;
import org.apache.http.util.EntityUtils;
import org.apache.log4j.Logger;
public class TestNodeAliveness {
private static Logger log = Logger.getLogger(TestNodeAliveness.class);
public static boolean nodeBIT(String elasticIP) throws ClientProtocolException, IOException {
try {
HttpClient client = new DefaultHttpClient();
// The time it takes to open TCP connection.
client.getParams().setParameter(CoreConnectionPNames.CONNECTION_TIMEOUT, 1);
// Timeout when server does not send data.
client.getParams().setParameter(CoreConnectionPNames.SO_TIMEOUT, 5000);
// Some tuning that is not required for bit tests.
client.getParams().setParameter(CoreConnectionPNames.STALE_CONNECTION_CHECK, false);
client.getParams().setParameter(CoreConnectionPNames.TCP_NODELAY, true);
HttpUriRequest request = new HttpGet("http://" + elasticIP);
HttpResponse response = client.execute(request);
HttpEntity entity = response.getEntity();
if(entity == null) {
return false;
} else {
System.out.println(EntityUtils.toString(entity));
}
// Close just in case.
request.abort();
} catch (Throwable e) {
log.warn("BIT Test failed for " + elasticIP);
e.printStackTrace();
return false;
}
return true;
}
public static void main(String[] args) throws ClientProtocolException, IOException {
nodeBIT("google.com?cant_cache_this=" + (new Random()).nextInt());
}
}
How is this possible?
Thank you.
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(2)
我使用过的所有 JVM 的有效超时粒度约为 15-30 毫秒。即使超时设置为 1 个套接字 I/O,如果连接请求花费的时间少于 15-30 毫秒,则通常会成功。
The effective granularity of time-outs in all JVMs I have worked with is approximately 15-30ms. Even if the timeout is set to 1 socket I/O and connect requests often succeed if they take less than 15-30ms.
如果你设置 HttpGet 的超时,而不是 HttpClient,它似乎工作得更好。
If you set the timeout of the HttpGet,instead of the HttpClient, it seems to work much better.