Spring Cloud组件那么多超时设置,我们如何理解和运用?_NPException的博客-CSDN博客_超时机制设计必须遵循什么原则前言Spring Cloud作为微服务解决方案 全家桶,集合了丰富的微服务组件,如Gateway、Feign、Hystrix,Ribbon、OkHttp、Eureka等等。而作为服务调用环节涉及到的几个组件:Feign、Hystrix,Ribbon、OkHttp都有超时时间的设置,Spring Cloud 是如何优雅地把它们协调好呢?Spring Cloud 在接口调用上,大致会经过如...https://blog.csdn.net/qq_36850813/article/details/100520361

Ribbon参数配置_chengqiuming的博客-CSDN博客_ribbon配置一 全局配置1 格式ribbon.&lt;key&gt;=&lt;value&gt;&lt;key&gt;代表了Ribbon客户端配置的参数名,&lt;value&gt;则代表了对应参数的值。2 举例ribbon.ConnectTimeout=2503 说明全局配置可以作为默认值进行设置,当指定客户端配置相应key值时,将覆盖全局配置的内容。二 指定客户端配置1...https://blog.csdn.net/chengqiuming/article/details/81318510CommonClientConfigKey 这个类包含了ribbon各种配置key,里面的key可以通过格式ribbon.<key>=<value>来设置它的值。比如,该类有一个变量为ConnectTimeout,则可以通过配置ribbon.ConnectTimeout=2000来设置连接超时时间。

如果是zuul(网关)的超时时间需要设置zuul、hystrix、ribbon等三部分:

zuul超时设置

#zuul超时设置
#默认1000
zuul.host.socket-timeout-millis=2000
#默认2000
zuul.host.connect-timeout-millis=4000

hystrix超时设置

#熔断器启用
feign.hystrix.enabled=true
hystrix.command.default.execution.timeout.enabled=true
#断路器的超时时间,下级服务返回超出熔断器时间,即便成功,消费端消息也是TIMEOUT,所以一般断路器的超时时间需要大于ribbon的超时时间,ribbon是真正去调用下级服务
#当服务的返回时间大于ribbon的超时时间,会触发重试
#断路器的超时时间默认为1000ms,太小了
hystrix.command.default.execution.isolation.thread.timeoutInMilliseconds=60000

#断路器详细设置
#当在配置时间窗口内达到此数量的失败后,进行短路。默认20个)
#hystrix.command.default.circuitBreaker.requestVolumeThreshold=20
#短路多久以后开始尝试是否恢复,默认5s)
#hystrix.command.default.circuitBreaker.sleepWindowInMilliseconds=5
#出错百分比阈值,当达到此阈值后,开始短路。默认50%)
#hystrix.command.default.circuitBreaker.errorThresholdPercentage=50%

ribbon超时设置

#ribbon请求连接的超时时间,限制3秒内必须请求到服务,并不限制服务处理的返回时间
ribbon.ConnectTimeout=3000

ribbon.SocketTimeout=5000
#请求处理的超时时间 下级服务响应最大时间,超出时间消费方(路由也是消费方)返回timeout
ribbon.ReadTimeout=5000
# 单独设置某个服务的超时时间,会覆盖其他的超时时间限制,服务的名称以注册中心页面显示的名称为准,超时时间不可大于断路器的超时时间
#service-a.ribbon.ReadTimeout=50000
#service-a.ribbon.ConnectTimeout=50000

重试机制

#重试机制
#该参数用来开启重试机制,默认是关闭
spring.cloud.loadbalancer.retry.enabled=true
#对所有操作请求都进行重试
ribbon.OkToRetryOnAllOperations=true
#对当前实例的重试次数
ribbon.MaxAutoRetries=1
#切换实例的重试次数
ribbon.MaxAutoRetriesNextServer=1
#根据如上配置,当访问到故障请求的时候,它会再尝试访问一次当前实例(次数由MaxAutoRetries配置),
#如果不行,就换一个实例进行访问,如果还是不行,再换一次实例访问(更换次数由MaxAutoRetriesNextServer配置),
#如果依然不行,返回失败信息。

按照上面的设置,应用会报警告,警告说hystrix的超时时间比ribbon的小,为什么会这样?

大家的实际在初次使用spring-cloud的gateway的时候,肯定会被里面各种的Timeout搞得晕头转向。hytrix有设置,ribbon也有。我们一开始也是乱设一桶,Github上各种项目里也没几个设置正确的。对Timeout的研究源于一次log中的warning

The Hystrix timeout of 60000 ms for the command “foo” is set lower than the combination of the Ribbon read and connect timeout, 200000ms.

hytrix超时时间

log出自AbstractRibbonCommand.java,那么索性研究一下源码。

假设:

这里gateway会请求一个serviceName=foo的服务
protected static int getHystrixTimeout(IClientConfig config, String commandKey) {

int ribbonTimeout = getRibbonTimeout(config, commandKey);

DynamicPropertyFactory dynamicPropertyFactory = DynamicPropertyFactory.getInstance();

// 获取默认的hytrix超时时间

int defaultHystrixTimeout = dynamicPropertyFactory.getIntProperty("hystrix.command.default.execution.isolation.thread.timeoutInMilliseconds",

0).get();

// 获取具体服务的hytrix超时时间,这里应该是hystrix.command.foo.execution.isolation.thread.timeoutInMilliseconds

int commandHystrixTimeout = dynamicPropertyFactory.getIntProperty("hystrix.command." + commandKey + ".execution.isolation.thread.timeoutInMilliseconds",

0).get();

int hystrixTimeout;

// hystrixTimeout的优先级是 具体服务的hytrix超时时间 > 默认的hytrix超时时间 > ribbon超时时间

if(commandHystrixTimeout > 0) {

hystrixTimeout = commandHystrixTimeout;

}

else if(defaultHystrixTimeout > 0) {

hystrixTimeout = defaultHystrixTimeout;

} else {

hystrixTimeout = ribbonTimeout;

}

// 如果默认的或者具体服务的hytrix超时时间小于ribbon超时时间就会警告

if(hystrixTimeout < ribbonTimeout) {

LOGGER.warn("The Hystrix timeout of " + hystrixTimeout + "ms for the command " + commandKey +

" is set lower than the combination of the Ribbon read and connect timeout, " + ribbonTimeout + "ms.");

}

return hystrixTimeout;

}

紧接着,看一下我们的配置是什么

hystrix:

command:

default:

execution:

isolation:

thread:

timeoutInMilliseconds: 60000

ribbon:

ReadTimeout: 50000

ConnectTimeout: 50000

MaxAutoRetries: 0

MaxAutoRetriesNextServer: 1

ribbon超时时间

这里ribbon的超时时间是50000ms,那么为什么log中写的ribbon时间是200000ms?

继续分析源码:

protected static int getRibbonTimeout(IClientConfig config, String commandKey) {

int ribbonTimeout;

// 这是比较异常的情况,不说

if (config == null) {

ribbonTimeout = RibbonClientConfiguration.DEFAULT_READ_TIMEOUT + RibbonClientConfiguration.DEFAULT_CONNECT_TIMEOUT;

} else {

// 这里获取了四个参数,ReadTimeout,ConnectTimeout,MaxAutoRetries, MaxAutoRetriesNextServer

int ribbonReadTimeout = getTimeout(config, commandKey, "ReadTimeout",

IClientConfigKey.Keys.ReadTimeout, RibbonClientConfiguration.DEFAULT_READ_TIMEOUT);

int ribbonConnectTimeout = getTimeout(config, commandKey, "ConnectTimeout",

IClientConfigKey.Keys.ConnectTimeout, RibbonClientConfiguration.DEFAULT_CONNECT_TIMEOUT);

int maxAutoRetries = getTimeout(config, commandKey, "MaxAutoRetries",

IClientConfigKey.Keys.MaxAutoRetries, DefaultClientConfigImpl.DEFAULT_MAX_AUTO_RETRIES);

int maxAutoRetriesNextServer = getTimeout(config, commandKey, "MaxAutoRetriesNextServer",

IClientConfigKey.Keys.MaxAutoRetriesNextServer, DefaultClientConfigImpl.DEFAULT_MAX_AUTO_RETRIES_NEXT_SERVER);

// 原来ribbonTimeout的计算方法在这里,以上文的设置为例

// ribbonTimeout = (50000 + 50000) * (0 + 1) * (1 + 1) = 200000

ribbonTimeout = (ribbonReadTimeout + ribbonConnectTimeout) * (maxAutoRetries + 1) * (maxAutoRetriesNextServer + 1);

}

return ribbonTimeout;

}

可以看到ribbonTimeout是一个总时间,所以从逻辑上来讲,作者希望hystrixTimeout要大于ribbonTimeout,否则hystrix熔断了以后,ribbon的重试就都没有意义了。

ribbon单服务设置

到这里最前面的疑问已经解开了,但是hytrix可以分服务设置timeout,ribbon可不可以? 源码走起,这里看的文件是DefaultClientConfigImpl.java

// 这是获取配置的入口方法,如果是null,那么用默认值

// 所有ribbon的默认值的都在该类中设置了,可以自己看一下

public <T> T get(IClientConfigKey<T> key, T defaultValue) {

T value = get(key);

if (value == null) {

value = defaultValue;

}

return value;

}

// 这是核心方法

protected Object getProperty(String key) {

if (enableDynamicProperties) {

String dynamicValue = null;

DynamicStringProperty dynamicProperty = dynamicProperties.get(key);

// dynamicProperties其实是一个缓存,首次访问foo服务的时候会加载

if (dynamicProperty != null) {

dynamicValue = dynamicProperty.get();

}

// 如果缓存没有,那么就再获取一次,注意这里的getConfigKey(key)是生成key的方法

if (dynamicValue == null) {

dynamicValue = DynamicProperty.getInstance(getConfigKey(key)).getString();

// 如果还是没有取默认值,getDefaultPropName(key)生成key的方法

if (dynamicValue == null) {

dynamicValue = DynamicProperty.getInstance(getDefaultPropName(key)).getString();

}

}

if (dynamicValue != null) {

return dynamicValue;

}

}

return properties.get(key);

}

以我们的服务为例:

getConfigKey(key) returns foo.ribbon.ReadTimeout

getDefaultPropName(key) returns ribbon.ReadTimeout

一目了然,{serviceName}.ribbon.{propertyName}就可以了。
 

上面的重试开关ribbon.OkToRetryOnAllOperations=true是对所有请求重试,但有时候我们不希望对某些请求重试,可以参考下面资料:

我们一般部署服务的时候,都会部署一个网关服务,内部所有的其他微服务的调用,都将通过网关路由过去,不对外直接暴露,对外只暴露网关服务。而且一般内部服务会部署多个实例,zuul集成了ribbon,会自动负载均衡的方式去调用内部服务。

当内部服务滚动重启的时候,通过网关的一个请求刚好路由到重启的那个实例的话,因为默认没有开启zuul的请求重试策略,该请求将会报错,其实理想的方式可以通过重试路由到另外一个活动的服务实例上去。

要开启zuul网关请求重试,首先需要添加spring-retry依赖:

        <dependency>
            <groupId>org.springframework.retry</groupId>
            <artifactId>spring-retry</artifactId>
        </dependency>

然后配置:

zuul.retryable=true
这样,所有路由都将会进行重试。(此属性默认是false,所以不会重试)

有时候我们不希望所有路由都有重试机制,我们可以配置指定路由进行重试:

zuul.routes.<routename>.retryable=true
这里的routename默认情况下就是你的服务名(我们可以通过管理端点/routes看到都有哪些路由,也可以查看更详细的路由信息:/routes?format=details,端点实现类:org.springframework.cloud.netflix.zuul.RoutesMvcEndpoint)。例如我有一个rcmd-service-data的服务,我可以这样配置:

zuul.retryable=false
zuul.routes.rcmd-service-data.retryable=true
这样,就只有rcmd-service-data这个服务开启了重试机制。我们通过/routes?format=details端点也可以看到:

我们知道zuul请求也是通过Ribbon负载均衡客户端去调用其他服务的,所以我们的重试策略也是在具体的ribbon配置中指定:

rcmd-service-data:
  ribbon:
    # Max number of retries on the same server (excluding the first try)
    MaxAutoRetries: 1 
    # Max number of next servers to retry (excluding the first server)
    MaxAutoRetriesNextServer: 2 #当允许在其他服务器上重试的时候,会调用IRule.choose选择可用服务实例中的其他一台服务实例进行调用
    # Whether all operations can be retried for this client
    OkToRetryOnAllOperations: true  #默认为false,则只允许GET请求被重试
    ReadTimeout: 5000
    ConnectTimeout: 2000
重试的时候还有补偿策略,例如重试时间间隔(默认是没有间隔:org.springframework.retry.backoff.NoBackOffPolicy),我们可以实现自己的补偿策略,也可以用内部实现的一些补偿策略(需要定义一个bean),如指数级的补偿策略(1秒,2秒,4秒类似这种指数级睡眠间隔增长,不超过10秒):
@Configuration
public class MyConfiguration {
    @Bean
    LoadBalancedBackOffPolicyFactory backOffPolciyFactory() {
        return new LoadBalancedBackOffPolicyFactory() {
            @Override
            public BackOffPolicy createBackOffPolicy(String service) {
                return new ExponentialBackOffPolicy();
            }
        };
    }
}
也可以正对某些响应状态码进行重试(当调用rcmd-service-data返回404,502的时候,进行重试,其他状态码不重试):

rcmd-service-data:
  ribbon:
    retryableStatusCodes: 404,502
以上差不多就是网关重试相关的能够配置的点了.

下面从源码层面看看重试的实现。

首先我们需要大致有个概念就是,zuul网关是通过ribbon负载均衡客户端来调用内部服务的,然后ribbon客户端默认是用HttpClient来发起http请求调用(当然还可以通过配置使用okhttp或者ribbon自带的RestClient),在org.springframework.cloud.netflix.ribbon.apache.HttpClientRibbonConfiguration中进行自动装配的:

@Bean
    @ConditionalOnMissingBean(AbstractLoadBalancerAwareClient.class)
    @ConditionalOnClass(name = "org.springframework.retry.support.RetryTemplate")
    public RetryableRibbonLoadBalancingHttpClient retryableRibbonLoadBalancingHttpClient(
        IClientConfig config, ServerIntrospector serverIntrospector,
        ILoadBalancer loadBalancer, RetryHandler retryHandler,
        LoadBalancedRetryPolicyFactory loadBalancedRetryPolicyFactory, CloseableHttpClient httpClient,
        LoadBalancedBackOffPolicyFactory loadBalancedBackOffPolicyFactory,
        LoadBalancedRetryListenerFactory loadBalancedRetryListenerFactory) {
        RetryableRibbonLoadBalancingHttpClient client = new RetryableRibbonLoadBalancingHttpClient(
            httpClient, config, serverIntrospector, loadBalancedRetryPolicyFactory,
            loadBalancedBackOffPolicyFactory, loadBalancedRetryListenerFactory);
        client.setLoadBalancer(loadBalancer);
        client.setRetryHandler(retryHandler);
        Monitors.registerObject("Client_" + this.name, client);
        return client;
    }
ribbon内部优惠构造hystrix command来执行请求(所以我们常说ribbon的超时时间设置要大于等于hystrix 超时时间,不然导致command还没执行完,ribbon却超时了),

由HttpClientRibbonCommand.run方法发起请求(org.springframework.cloud.netflix.zuul.filters.route.support.AbstractRibbonCommand#run):

    @Override
    protected ClientHttpResponse run() throws Exception {
        final RequestContext context = RequestContext.getCurrentContext();
 
        RQ request = createRequest();
        RS response;
        
        boolean retryableClient = this.client instanceof AbstractLoadBalancingClient
                && ((AbstractLoadBalancingClient)this.client).isClientRetryable((ContextAwareRequest)request);
        
        if (retryableClient) {
            response = this.client.execute(request, config); //此处的client就是
RetryableRibbonLoadBalancingHttpClient
} else {response = this.client.executeWithLoadBalancer(request, config);}context.set("ribbonResponse", response);// Explicitly close the HttpResponse if the Hystrix command timed out to// release the underlying HTTP connection held by the response.//if (this.isResponseTimedOut()) {if (response != null) {response.close();}}return new RibbonHttpResponse(response);}
如果可以重试的话,调用的是RetryableRibbonLoadBalancingHttpClient.execute方法:

@Override
    public RibbonApacheHttpResponse execute(final RibbonApacheHttpRequest request, final IClientConfig configOverride) throws Exception {
        final RequestConfig.Builder builder = RequestConfig.custom();
        IClientConfig config = configOverride != null ? configOverride : this.config;
        builder.setConnectTimeout(config.get(
                CommonClientConfigKey.ConnectTimeout, this.connectTimeout));
        builder.setSocketTimeout(config.get(
                CommonClientConfigKey.ReadTimeout, this.readTimeout));
        builder.setRedirectsEnabled(config.get(
                CommonClientConfigKey.FollowRedirects, this.followRedirects));
 
        final RequestConfig requestConfig = builder.build();
        final LoadBalancedRetryPolicy retryPolicy = loadBalancedRetryPolicyFactory.create(this.getClientName(), this);
        RetryCallback<RibbonApacheHttpResponse, IOException> retryCallback = new RetryCallback<RibbonApacheHttpResponse, IOException>() {
            @Override
            public RibbonApacheHttpResponse doWithRetry(RetryContext context) throws IOException {
                //on retries the policy will choose the server and set it in the context
                //extract the server and update the request being made
                RibbonApacheHttpRequest newRequest = request;
                if(context instanceof LoadBalancedRetryContext) {
                    ServiceInstance service = ((LoadBalancedRetryContext)context).getServiceInstance();
                    if(service != null) {
                        //Reconstruct the request URI using the host and port set in the retry context
                        newRequest = newRequest.withNewUri(UriComponentsBuilder.newInstance().host(service.getHost())
                                .scheme(service.getUri().getScheme()).userInfo(newRequest.getURI().getUserInfo())
                                .port(service.getPort()).path(newRequest.getURI().getPath())
                                .query(newRequest.getURI().getQuery()).fragment(newRequest.getURI().getFragment())
                                .build().encode().toUri());
                    }
                }
                newRequest = getSecureRequest(newRequest, configOverride);
                HttpUriRequest httpUriRequest = newRequest.toRequest(requestConfig);
                final HttpResponse httpResponse = RetryableRibbonLoadBalancingHttpClient.this.delegate.execute(httpUriRequest);
                if(retryPolicy.retryableStatusCode(httpResponse.getStatusLine().getStatusCode())) {
                    throw new HttpClientStatusCodeException(RetryableRibbonLoadBalancingHttpClient.this.clientName,
                            httpResponse, HttpClientUtils.createEntity(httpResponse), httpUriRequest.getURI());
                }
                return new RibbonApacheHttpResponse(httpResponse, httpUriRequest.getURI());
            }
        };
        RibbonRecoveryCallback<RibbonApacheHttpResponse, HttpResponse> recoveryCallback = new RibbonRecoveryCallback<RibbonApacheHttpResponse, HttpResponse>() {
            @Override
            protected RibbonApacheHttpResponse createResponse(HttpResponse response, URI uri) {
                return new RibbonApacheHttpResponse(response, uri);
            }
         };
        return this.executeWithRetry(request, retryPolicy, retryCallback, recoveryCallback);
    }
    
    
进而又调用了自身的executeWithRetry方法:

private RibbonApacheHttpResponse executeWithRetry(RibbonApacheHttpRequest request, LoadBalancedRetryPolicy retryPolicy,
                                      RetryCallback<RibbonApacheHttpResponse, IOException> callback,
                                      RecoveryCallback<RibbonApacheHttpResponse> recoveryCallback) throws Exception {
   RetryTemplate retryTemplate = new RetryTemplate();
   boolean retryable = isRequestRetryable(request); //从org.springframework.cloud.netflix.zuul.filters.route.RibbonCommandContext中获取retryable配置
   retryTemplate.setRetryPolicy(retryPolicy == null || !retryable ? new NeverRetryPolicy()
         : new RetryPolicy(request, retryPolicy, this, this.getClientName()));
   BackOffPolicy backOffPolicy = loadBalancedBackOffPolicyFactory.createBackOffPolicy(this.getClientName());
   retryTemplate.setBackOffPolicy(backOffPolicy == null ? new NoBackOffPolicy() : backOffPolicy);
   RetryListener[] retryListeners = this.loadBalancedRetryListenerFactory.createRetryListeners(this.getClientName());
   if (retryListeners != null && retryListeners.length != 0) {
      retryTemplate.setListeners(retryListeners);
   }
   return retryTemplate.execute(callback, recoveryCallback);
}
private boolean isRequestRetryable(ContextAwareRequest request) {
   return request.getContext() == null ? true :
      BooleanUtils.toBooleanDefaultIfNull(request.getContext().getRetryable(), true);
}
executeWithRetry方法又会调回execute方法中定义的RetryCallback匿名实现类的doWithRetry方法:

RetryCallback<RibbonApacheHttpResponse, IOException> retryCallback = new RetryCallback<RibbonApacheHttpResponse, IOException>() {
            @Override
            public RibbonApacheHttpResponse doWithRetry(RetryContext context) throws IOException {
                //on retries the policy will choose the server and set it in the context
                //extract the server and update the request being made
                RibbonApacheHttpRequest newRequest = request;
                if(context instanceof LoadBalancedRetryContext) {
                    ServiceInstance service = ((LoadBalancedRetryContext)context).getServiceInstance();//这个getServiceInstance在next server retry的时候会变
                    if(service != null) {
                        //Reconstruct the request URI using the host and port set in the retry context
                        newRequest = newRequest.withNewUri(UriComponentsBuilder.newInstance().host(service.getHost())
                                .scheme(service.getUri().getScheme()).userInfo(newRequest.getURI().getUserInfo())
                                .port(service.getPort()).path(newRequest.getURI().getPath())
                                .query(newRequest.getURI().getQuery()).fragment(newRequest.getURI().getFragment())
                                .build().encode().toUri());
                    }
                }
                newRequest = getSecureRequest(newRequest, configOverride);
                HttpUriRequest httpUriRequest = newRequest.toRequest(requestConfig);
                final HttpResponse httpResponse = RetryableRibbonLoadBalancingHttpClient.this.delegate.execute(httpUriRequest);
                if(retryPolicy.retryableStatusCode(httpResponse.getStatusLine().getStatusCode())) { //这里就是跟配置中配置的状态码进行匹配
                    throw new HttpClientStatusCodeException(RetryableRibbonLoadBalancingHttpClient.this.clientName,
                            httpResponse, HttpClientUtils.createEntity(httpResponse), httpUriRequest.getURI());
                }
                return new RibbonApacheHttpResponse(httpResponse, httpUriRequest.getURI());
            }
        };
我们需要缕一缕这部分代码逻辑,首先是这句:

boolean retryable = isRequestRetryable(request); //从org.springframework.cloud.netflix.zuul.filters.route.RibbonCommandContext中获取retryable配置
表名了从org.springframework.cloud.netflix.zuul.filters.route.RibbonCommandContext获取到的是否可以重试的配置,而
RibbonCommandContext又是从:

org.springframework.cloud.netflix.zuul.filters.route.RibbonRoutingFilter#buildCommandContext中创建出来的:

    protected RibbonCommandContext buildCommandContext(RequestContext context) {
        HttpServletRequest request = context.getRequest();
 
        MultiValueMap<String, String> headers = this.helper
                .buildZuulRequestHeaders(request);
        MultiValueMap<String, String> params = this.helper
                .buildZuulRequestQueryParams(request);
        String verb = getVerb(request);
        InputStream requestEntity = getRequestBody(request);
        if (request.getContentLength() < 0 && !verb.equalsIgnoreCase("GET")) {
            context.setChunkedRequestBody();
        }
 
        String serviceId = (String) context.get(SERVICE_ID_KEY);
        Boolean retryable = (Boolean) context.get(RETRYABLE_KEY);
        Object loadBalancerKey = context.get(LOAD_BALANCER_KEY);
 
        String uri = this.helper.buildZuulRequestURI(request);
 
        // remove double slashes
        uri = uri.replace("//", "/");
 
        long contentLength = useServlet31 ? request.getContentLengthLong(): request.getContentLength();
 
        return new RibbonCommandContext(serviceId, verb, uri, retryable, headers, params,
                requestEntity, this.requestCustomizers, contentLength, loadBalancerKey);
    }
这句:
Boolean retryable = (Boolean) context.get(RETRYABLE_KEY);
retryable又是从:

com.netflix.zuul.context.RequestContext中获取的,那么从RequestContext里面的这个key又是从哪设置进去的呢?我们进而搜索发现在:

org.springframework.cloud.netflix.zuul.filters.pre.PreDecorationFilter

这个zuul filter中设置的该值,run方法代码片段:

Route route = this.routeLocator.getMatchingRoute(requestURI);
if (route.getRetryable() != null) {
                    ctx.put(RETRYABLE_KEY, route.getRetryable());
                }
此处的routeLocator是:org.springframework.cloud.netflix.zuul.filters.CompositeRouteLocator.

这里用了一种设计模式(组合模式),不过里面就一个locator:

org.springframework.cloud.netflix.zuul.filters.discovery.DiscoveryClientRouteLocator,继承自:

org.springframework.cloud.netflix.zuul.filters.SimpleRouteLocator

最终会调用到:

org.springframework.cloud.netflix.zuul.filters.SimpleRouteLocator#getRoute方法:

protected Route getRoute(ZuulRoute route, String path) {
        if (route == null) {
            return null;
        }
        if (log.isDebugEnabled()) {
            log.debug("route matched=" + route);
        }
        String targetPath = path;
        String prefix = this.properties.getPrefix();
        if(prefix.endsWith("/")) {
            prefix = prefix.substring(0, prefix.length() - 1);
        }
        if (path.startsWith(prefix + "/") && this.properties.isStripPrefix()) {
            targetPath = path.substring(prefix.length());
        }
        if (route.isStripPrefix()) {
            int index = route.getPath().indexOf("*") - 1;
            if (index > 0) {
                String routePrefix = route.getPath().substring(0, index);
                targetPath = targetPath.replaceFirst(routePrefix, "");
                prefix = prefix + routePrefix;
            }
        }
        Boolean retryable = this.properties.getRetryable();
        if (route.getRetryable() != null) {
            retryable = route.getRetryable();
        }
        return new Route(route.getId(), targetPath, route.getLocation(), prefix,
                retryable,
                route.isCustomSensitiveHeaders() ? route.getSensitiveHeaders() : null, 
                route.isStripPrefix());
    }
看到这句:
Boolean retryable = this.properties.getRetryable();
最终取的就是本文开头的配置文件中的配置属性。

看到这里,我们终于把retryable属性的来龙去脉搞清楚了,下面说说怎么重试到另外一个实例上去的。

我们在看回:

org.springframework.cloud.netflix.ribbon.apache.RetryableRibbonLoadBalancingHttpClient#execute方法。

其中有一段代码:

RetryCallback<RibbonApacheHttpResponse, IOException> retryCallback = new RetryCallback<RibbonApacheHttpResponse, IOException>() {
   @Override
   public RibbonApacheHttpResponse doWithRetry(RetryContext context) throws IOException {
      //on retries the policy will choose the server and set it in the context
      //extract the server and update the request being made
      RibbonApacheHttpRequest newRequest = request;
      if(context instanceof LoadBalancedRetryContext) {
         ServiceInstance service = ((LoadBalancedRetryContext)context).getServiceInstance();
         if(service != null) {
            //Reconstruct the request URI using the host and port set in the retry context
            newRequest = newRequest.withNewUri(UriComponentsBuilder.newInstance().host(service.getHost())
                  .scheme(service.getUri().getScheme()).userInfo(newRequest.getURI().getUserInfo())
                  .port(service.getPort()).path(newRequest.getURI().getPath())
                  .query(newRequest.getURI().getQuery()).fragment(newRequest.getURI().getFragment())
                  .build().encode().toUri());
         }
      }
      newRequest = getSecureRequest(newRequest, configOverride);
      HttpUriRequest httpUriRequest = newRequest.toRequest(requestConfig);
      final HttpResponse httpResponse = RetryableRibbonLoadBalancingHttpClient.this.delegate.execute(httpUriRequest);
      if(retryPolicy.retryableStatusCode(httpResponse.getStatusLine().getStatusCode())) {
         throw new HttpClientStatusCodeException(RetryableRibbonLoadBalancingHttpClient.this.clientName,
               httpResponse, HttpClientUtils.createEntity(httpResponse), httpUriRequest.getURI());
      }
      return new RibbonApacheHttpResponse(httpResponse, httpUriRequest.getURI());
   }
};
前面已经提到过:

 ServiceInstance service = ((LoadBalancedRetryContext)context).getServiceInstance();
这一行代码getServiceInstance可能会变,即跟你最开始的路由不同,比如我有A,B两个实例,最开始的时候确定路由到A实例上,当A挂了,再重试的时候可能这个ServiceInstance可能是 Instance B。既然有get那么肯定有set它的地方.前面我们说了要实现重试,必须引入spring-retry依赖,然后才会通过org.springframework.retry.support.RetryTemplate#doExecute来执行重试请求,我们看看这个方法的代码片段:

while (canRetry(retryPolicy, context) && !context.isExhaustedOnly()) {

   try {
      if (this.logger.isDebugEnabled()) {
         this.logger.debug("Retry: count=" + context.getRetryCount());
      }
      // Reset the last exception, so if we are successful
      // the close interceptors will not think we failed...
      lastException = null;
      return retryCallback.doWithRetry(context); //这里调用了RetryableRibbonLoadBalancingHttpClient#execute中定义的RetryCallback的doWithRetry方法
   }
   catch (Throwable e) {

      lastException = e;

      try {
         registerThrowable(retryPolicy, state, context, e);//在失败的时候执行的逻辑,这里是关键,里面有重新选择服务实例的逻辑
      }
      catch (Exception ex) {
         throw new TerminatedRetryException("Could not register throwable",
               ex);
      }
      finally {
         doOnErrorInterceptors(retryCallback, context, e);
      }

      if (canRetry(retryPolicy, context) && !context.isExhaustedOnly()) {
         try {
            backOffPolicy.backOff(backOffContext);
         }
         catch (BackOffInterruptedException ex) {
            lastException = e;
            // back off was prevented by another thread - fail the retry
            if (this.logger.isDebugEnabled()) {
               this.logger
                     .debug("Abort retry because interrupted: count="
                           + context.getRetryCount());
            }
            throw ex;
         }
      }

      if (this.logger.isDebugEnabled()) {
         this.logger.debug(
               "Checking for rethrow: count=" + context.getRetryCount());
      }

      if (shouldRethrow(retryPolicy, context, state)) {
         if (this.logger.isDebugEnabled()) {
            this.logger.debug("Rethrow in retry for policy: count="
                  + context.getRetryCount());
         }
         throw RetryTemplate.<E>wrapIfNecessary(e);
      }

   }

   /*
    * A stateful attempt that can retry may rethrow the exception before now,
    * but if we get this far in a stateful retry there's a reason for it,
    * like a circuit breaker or a rollback classifier.
    */
   if (state != null && context.hasAttribute(GLOBAL_STATE)) {
      break;
   }
}
这行代码是关键,在请求发生异常的时候catch住异常,然后会执行:

registerThrowable(retryPolicy, state, context, e)
其中的retryPolicy是在RetryableRibbonLoadBalancingHttpClient中定义的内部类:
org.springframework.cloud.netflix.ribbon.apache.RetryableRibbonLoadBalancingHttpClient.RetryPolicy,所以其实调用委托给了:

org.springframework.cloud.netflix.ribbon.RibbonLoadBalancedRetryPolicy#registerThrowable方法:

@Override
public void registerThrowable(LoadBalancedRetryContext context, Throwable throwable) {
   //if this is a circuit tripping exception then notify the load balancer
   if (lbContext.getRetryHandler().isCircuitTrippingException(throwable)) {
      updateServerInstanceStats(context);
   }
   
   //Check if we need to ask the load balancer for a new server.
   //Do this before we increment the counters because the first call to this method
   //is not a retry it is just an initial failure.
   if(!canRetrySameServer(context)  && canRetryNextServer(context)) {
      context.setServiceInstance(loadBalanceChooser.choose(serviceId));
   }
   //This method is called regardless of whether we are retrying or making the first request.
   //Since we do not count the initial request in the retry count we don't reset the counter
   //until we actually equal the same server count limit.  This will allow us to make the initial
   //request plus the right number of retries.
   if(sameServerCount >= lbContext.getRetryHandler().getMaxRetriesOnSameServer() && canRetry(context)) {
      //reset same server since we are moving to a new server
      sameServerCount = 0;
      nextServerCount++;
      if(!canRetryNextServer(context)) {
         context.setExhaustedOnly();
      }
   } else {
      sameServerCount++;
   }

}
这个类中就用到了最开始说到的MaxAutoRetries,MaxAutoRetriesNextServer,OkToRetryOnAllOperations这三个属性。
主要看这段:

   if(!canRetrySameServer(context)  && canRetryNextServer(context)) {
      context.setServiceInstance(loadBalanceChooser.choose(serviceId));
   }
当能在其他服务器上重试的时候,然后通过loadBalanceChooser.choose(serviceId)重新选择了另外一个服务实例。然后set进了context。

整个流程大致上就是这样,具体的实现细节可以自行深入查看一下以上提及的类的代码。

另外还有一个注意事项,此源码是基于Edgware.SR2版本,在Edgware.SR1的时候有BUG,在:

org.springframework.cloud.netflix.ribbon.apache.RetryableRibbonLoadBalancingHttpClient#execute中定义的匿名类RetryCallback中这行代码

newRequest = getSecureRequest(newRequest, configOverride);
在SR1版本的时候是:

newRequest = getSecureRequest(request, configOverride);
导致retry next server无效,每次都被设置回了最开始route的实例,我也是在debug的时候发现的,开始用的SR1版本,我说怎么retry老是无效。。。后面去翻了一下官方的issue,发现:

https://github.com/spring-cloud/spring-cloud-netflix/issues/2667

此BUG在spring-cloud-netfix 1.4.3修复,SR1用的是1.4.2,SR2用的是1.4.3.

小伙伴们发现有问题,请确认你的版本是否和我说的版本一致,根据需要升级一下spring cloud版本。。。
 

转自:spring cloud各种超时时间设置

Hystrix超时时间与Ribbon超时时间关系

spring cloud zuul网关服务重试请求配置和源码分析

Logo

为开发者提供学习成长、分享交流、生态实践、资源工具等服务,帮助开发者快速成长。

更多推荐