概述
1. 概述
在默认情况下 spring cloud feign在进行各个子服务之间的调用时,http组件使用的是jdk的HttpURLConnection,没有使用线程池。本文先从源码分析feign的http组件对象生成的过程,然后通过为feign配置http线程池优化调用效率。
2. 源码分析
我们分析源码spring cloud feign。在spring-cloud-netflix-core/META-INF/spring.factories中可以看到,在spring boot自动配置会初始化FeignRibbonClientAutoConfiguration,这个类会生成Ribbon的使用http组件。
org.springframework.boot.autoconfigure.EnableAutoConfiguration=
org.springframework.cloud.netflix.feign.ribbon.FeignRibbonClientAutoConfiguration,
分析配置类是FeignRibbonClientAutoConfiguration 下面分析此类import的3个类:HttpClientFeignLoadBalancedConfiguration,OkHttpFeignLoadBalancedConfiguration,DefaultFeignLoadBalancedConfiguration
@Import({ HttpClientFeignLoadBalancedConfiguration.class,
OkHttpFeignLoadBalancedConfiguration.class,
DefaultFeignLoadBalancedConfiguration.class })
public class FeignRibbonClientAutoConfiguration {
…
}
HttpClientFeignLoadBalancedConfiguration
为feigin配置appache client的线程池
当引入ApacheHttpClient.class类时,会初始化这个配置类
方法feignClient()中:根据@ConditionalOnMissingBean(Client.class)知道如果有HttpClient 对象,则创建的ApacheHttpClient使用自己定义的HttpClient 。如果没有,则使用默认值。最后生成LoadBalancerFeignClient对象
@Configuration
@ConditionalOnClass(ApacheHttpClient.class)
@ConditionalOnProperty(value = "feign.httpclient.enabled", matchIfMissing = true)
class HttpClientFeignLoadBalancedConfiguration {
@Autowired(required = false)
private HttpClient httpClient;
@Bean
@ConditionalOnMissingBean(Client.class)
public Client feignClient(CachingSpringLoadBalancerFactory cachingFactory,
SpringClientFactory clientFactory) {
ApacheHttpClient delegate;
if (this.httpClient != null) {
delegate = new ApacheHttpClient(this.httpClient);
} else {
delegate = new ApacheHttpClient();
}
return new LoadBalancerFeignClient(delegate, cachingFactory, clientFactory);
}
}
OkHttpFeignLoadBalancedConfiguration
为feigin配置OkHttp,类似apache httpclient, 这里略。
DefaultFeignLoadBalancedConfiguration
为feigin配置HttpURLConnection,
方法feignClient():只有以上两个Client没有生产对象时,才在这个方法中使用Client.Default生成LoadBalancerFeignClient
@Configuration
class DefaultFeignLoadBalancedConfiguration {
@Bean
@ConditionalOnMissingBean
public Client feignClient(CachingSpringLoadBalancerFactory cachingFactory,
SpringClientFactory clientFactory) {
return new LoadBalancerFeignClient(new Client.Default(null, null),
cachingFactory, clientFactory);
}
}
查看Client.Default的源码,Default 使用HttpURLConnection 建立连接且每次请求都建立一个新的连接
public static class Default implements Client {
@Override
public Response execute(Request request, Options options) throws IOException {
HttpURLConnection connection = convertAndSend(request, options);
return convertResponse(connection).toBuilder().request(request).build();
}
….
}
综上所述,在默认情况下,spring cloud 没有引入httpclient和okhttp的jar包,所有默认使用HttpURLConnection
3. 使用appach httpclient线程池
默认情况下,服务之间调用使用的HttpURLConnection,效率非常低。为了提高效率,可以通过连接池提高效率,本节我们使用appache httpclient做为连接池。配置OkHttpClient连接池,也是类似的方法,这里略。 经过上节的分析,配置线程池方法:引入appache httpclient并启动对应配置,最后还需要生成HttpClient对象。
3.1. pom.xml中引入feign-httpclient.jar
<!-- 增加feign-httpclient -->
<dependency>
<groupId>io.github.openfeign</groupId>
<artifactId>feign-httpclient</artifactId>
</dependency>
3.2. 配置参数application-hystrix-feign.yml启动httpclient
# feign配置
feign:
hystrix:
# 在feign中开启hystrix功能,默认情况下feign不开启hystrix功能
enabled: true
## 配置httpclient线程池
httpclient:
enabled: true
okhttp:
enabled: false
3.3. 自定义配置类
使用配置类,生成HttpClient 对象。因为使用PoolingHttpClientConnectionManager连接池,我们需要启动定时器,定时回收过期的连接。配置定时回收连接池的原因,见问题备忘: httpclient连接池异常引发的惨案
@Configuration
public class HttpPool {
@Bean
public HttpClient httpClient(){
System.out.println("init feign httpclient configuration " );
// 生成默认请求配置
RequestConfig.Builder requestConfigBuilder = RequestConfig.custom();
// 超时时间
requestConfigBuilder.setSocketTimeout(5 * 1000);
// 连接时间
requestConfigBuilder.setConnectTimeout(5 * 1000);
RequestConfig defaultRequestConfig = requestConfigBuilder.build();
// 连接池配置
// 长连接保持30秒
final PoolingHttpClientConnectionManager pollingConnectionManager = new PoolingHttpClientConnectionManager(30, TimeUnit.MILLISECONDS);
// 总连接数
pollingConnectionManager.setMaxTotal(5000);
// 同路由的并发数
pollingConnectionManager.setDefaultMaxPerRoute(100);
// httpclient 配置
HttpClientBuilder httpClientBuilder = HttpClientBuilder.create();
// 保持长连接配置,需要在头添加Keep-Alive
httpClientBuilder.setKeepAliveStrategy(new DefaultConnectionKeepAliveStrategy());
httpClientBuilder.setConnectionManager(pollingConnectionManager);
httpClientBuilder.setDefaultRequestConfig(defaultRequestConfig);
HttpClient client = httpClientBuilder.build();
// 启动定时器,定时回收过期的连接
Timer timer = new Timer();
timer.schedule(new TimerTask() {
@Override
public void run() {
//
System.out.println("=====closeIdleConnections===");
pollingConnectionManager.closeExpiredConnections();
pollingConnectionManager.closeIdleConnections(5, TimeUnit.SECONDS);
}
}, 10 * 1000, 5 * 1000);
System.out.println("===== Apache httpclient 初始化连接池===");
return client;
}
}
3.4. 测试:
启动工程:cloud-registration-center、cloud-service-hystrix
启动服务:HystrixFeignCloudConsumerApplication
执行请求:http://127.0.0.1:12082/hystrix-feign/simple
配置日志为debug输出(设置logback-spring.xml为 level为DEBUG),如果日志有类似一下的输出(包含PoolingHttpClientConnectionManager ),则表示连接池配置成功
2018-04-09 23:11:49.017 [hystrix-cloud-hystrix-service-1] DEBUG o.a.h.i.c.PoolingHttpClientConnectionManager - Connection request: [route: {}->http://192.168.0.113:12081][total kept alive: 0; route allocated: 0 of 100; total allocated: 0 of 5000]
2018-04-09 23:11:49.020 [hystrix-cloud-hystrix-service-1] DEBUG o.a.h.i.c.PoolingHttpClientConnectionManager - Connection leased: [id: 0][route: {}->http://192.168.0.113:12081][total kept alive: 0; route allocated: 1 of 100; total allocated: 1 of 5000]
4. 代码
以上的详细的代码见下面
github代码,请尽量使用tag v0.12
最后
以上就是大力口红为你收集整理的【转】Spring cloud feign 使用线程池的全部内容,希望文章能够帮你解决【转】Spring cloud feign 使用线程池所遇到的程序开发问题。
如果觉得靠谱客网站的内容还不错,欢迎将靠谱客网站推荐给程序员好友。
发表评论 取消回复