SpringCloud Sleuth 简介
Spring Cloud Sleuth为Spring Cloud实现了分布式跟踪解决方案。
Spring Cloud Sleuth借鉴了Dapper的术语。
Span:基本的工作单元。Span包括一个64位的唯一ID,一个64位trace码,描述信息,时间戳事件,key-value 注解(tags),span处理者的ID(通常为IP)。
Trace:一组Span形成的树形结构。
Annotation:用于及时记录存在的事件。常用的Annotation如下:
- cs:客户端发送(client send) 客户端发起一个请求,表示span开始
- sr:服务器接收(server received) 服务器接收到客户端的请求并开始处理,sr - cs 的时间为网络延迟
- ss:服务器发送(server send) 服务器处理完请求准备返回数据给客户端。ss - sr 的时间表示服务器端处理请求花费的时间
- cr:客户端接收(client received) 客户端接收到处理结果,表示span结束。 cr - cs 的时间表示客户端接收服务端数据的时间
下图展示了Span和Trace在系统中的联系
SpringCloud Sleuth 默认采用 Http 方式将 span 信息传输给 Zipkin
在application.properties文件中指定
spring.zipkin.sender.type=web
使用 RabbitMQ 异步发送 span 信息
为什么选择 RabbitMQ 消息中间件发送 span 信息
- sleuth 默认采用 http 通信方式,将数据传给 zipkin 作页面渲染,但是 http 传输过程中如果由于不可抗因素导致 http 通信中断,那么此次通信的数据将会丢失。而使用中间件的话,RabbitMQ 消息队列可以积压千万级别的消息,下次重连之后可以继续消费。
- 随着线程增多,并发量提升之后,RabbitMQ 异步发送数据明显更具有优势。
- RabbitMQ 支持消息、队列持久化,可以通过消息状态落库、重回队列、镜像队列等技术手段保证其高可用。
示例
示例简介
示例包含sleuth-search、sleuth-cart、sleuth-order三个系统,用来模拟电商系统中下单的流程,用户可以搜索商品然后立即下单,也可以搜索多个商品后加入购物车,然后下单,调用情况即 search -> cart -> order,或 search -> order。
示例使用 RestTemplate
来完成三个系统间的 http
请求响应,请求方式也都遵循Restful
风格。
版本说明
版本一定要对应好,一些低版本的SpringBoot无法兼容新版本的SpringCloud和zipkin
工具 | 版本 |
---|---|
SpringBoot | 2.1.6.RELEASE |
SpringCloud | Greenwich.SR3 |
zipkin | 2.16.2 |
项目结构
demo-cloudsleuth
|- sleuth-search
|- sleuth-cart
|- sleuth-order
pom.xml
导入依赖
<!-- 引入 springboot 和 springcloud 父工程 -->
<parent>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-starter-parent</artifactId>
<version>2.1.6.RELEASE</version>
<relativePath/>
</parent>
<dependencyManagement>
<dependencies>
<dependency>
<groupId>org.springframework.cloud</groupId>
<artifactId>spring-cloud-dependencies</artifactId>
<version>Greenwich.SR3</version>
<type>pom</type>
<scope>import</scope>
</dependency>
</dependencies>
</dependencyManagement>
<dependencies>
<dependency>
<groupId>org.springframework.cloud</groupId>
<artifactId>spring-cloud-starter-zipkin</artifactId>
</dependency>
<dependency>
<groupId>org.springframework.amqp</groupId>
<artifactId>spring-rabbit</artifactId>
</dependency>
<!-- Springboot 相关 -->
<dependency>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-starter-web</artifactId>
</dependency>
<dependency>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-starter-test</artifactId>
<scope>test</scope>
</dependency>
</dependencies>
配置 RestTemplate,RestTemplate是SpringBoot提供的封装好的http工具类,可以帮助我们简化http的使用。
package com.anqi.cart.resttmplate;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.http.client.ClientHttpRequestFactory;
import org.springframework.http.client.SimpleClientHttpRequestFactory;
import org.springframework.web.client.RestTemplate;
@Configuration
public class RestTemplateConfig {
@Bean
public RestTemplate restTemplate(ClientHttpRequestFactory factory) {
return new RestTemplate(factory);
}
@Bean
public ClientHttpRequestFactory clientHttpRequestFactory() {
SimpleClientHttpRequestFactory factory = new SimpleClientHttpRequestFactory();
factory.setConnectTimeout(5000);
factory.setReadTimeout(5000);
return factory;
}
}
三个系统下的application.properties,端口分别是8081 8082 8083
#server.port=8081 server.port=8082
server.port=8083
server.servlet.context-path=/
spring.zipkin.base-url=http://localhost:9411/
spring.zipkin.service.name=sleuth-cart
#使用默认 http 方式收集 span 需要配置此项
#spring.zipkin.sender.type=web
#sleuth 使用 rabbitmq 来向 zipkin 发送数据
spring.zipkin.sender.type=rabbit
spring.rabbitmq.host=localhost
spring.rabbitmq.port=5672
spring.rabbitmq.username=guest
spring.rabbitmq.password=guest
#设置采样率默认为 0.1 注意之前的版本是percentage 新版本中更换为 probability
spring.sleuth.sampler.probability=1
三个系统下的RestTemplate的配置,用来简化 http 请求
package com.anqi.cart.resttmplate;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.http.client.ClientHttpRequestFactory;
import org.springframework.http.client.SimpleClientHttpRequestFactory;
import org.springframework.web.client.RestTemplate;
@Configuration
public class RestTemplateConfig {
@Bean
public RestTemplate restTemplate(ClientHttpRequestFactory factory) {
return new RestTemplate(factory);
}
@Bean
public ClientHttpRequestFactory clientHttpRequestFactory() {
SimpleClientHttpRequestFactory factory = new SimpleClientHttpRequestFactory();
factory.setConnectTimeout(5000);
factory.setReadTimeout(5000);
return factory;
}
}
@RequestMapping("cart")
@RestController
public class CartController {
@Autowired
RestTemplate restTemplate;
@Autowired
CartService cartService;
private static final String orderUrl = "http://localhost:8084/order/create";
@GetMapping("/add/{cartId}")
public String addToCart(@PathVariable("cartId") String cartId) {
cartService.addProductToCart(cartId, "小米8");
ResponseEntity<String> res = restTemplate.getForEntity(orderUrl, String.class);
return res.getBody();
}
}
@RequestMapping("order")
@RestController
public class OrderController {
@GetMapping("/create")
public String creatOrder() {
System.out.println("create order");
return "create_order";
}
}
@RestController
public class SearchController {
@Autowired
RestTemplate restTemplate;
private static final String cartUrl = "http://localhost:8083/cart/add/1";
private static final String orderUrl = "http://localhost:8084/order/create";
@GetMapping("/search")
public String search() {
ResponseEntity<String> cartRes = restTemplate.getForEntity(cartUrl, String.class);
ResponseEntity<String> orderRes = restTemplate.getForEntity(orderUrl, String.class);
return "cart:" + cartRes.getBody() + "- order:" + orderRes.getBody();
}
}
运行结果分析
默认 http 传输 span 信息
启动Zipkin
java -jar zipkin-server-2.16.2-exec.jar
网页中手动访问
http://localhost:8082/search
我们访问zipkin站点查询调用情况
http://localhost:9411/zipkin/traces/94b954d843012ca9
可以从下图中完整清晰的看到三个系统的调用关系
下图为zipkin调用预览,我们请求四次http://localhost:8082/search
来更直观的观察数据。在以下界面中,较为简洁的显示Span的个数以及调用总时延。
我们进入一个完整的调用链后访问其中的一个节点得到以下数据。
以下为一次全链路追踪的详细信息,包含7个span的所有信息,以上看到的页面展示均有以下数据加以渲染而成。
[
{
"traceId": "94b954d843012ca9",
"parentId": "bab70b1e69a5f3e3",
"id": "96387b33a823ca8f",
"kind": "SERVER",
"name": "get /order/create",
"timestamp": 1569060494069123,
"duration": 1161,
"localEndpoint": {
"serviceName": "sletuth-order",
"ipv4": "192.168.0.107"
},
"remoteEndpoint": {
"ipv4": "127.0.0.1",
"port": 49863
},
"tags": {
"http.method": "GET",
"http.path": "/order/create",
"mvc.controller.class": "OrderController",
"mvc.controller.method": "creatOrder"
},
"shared": true
},
{
"traceId": "94b954d843012ca9",
"parentId": "94b954d843012ca9",
"id": "90f7e5cfa89e0d80",
"kind": "SERVER",
"name": "get /order/create",
"timestamp": 1569060494076287,
"duration": 1296,
"localEndpoint": {
"serviceName": "sletuth-order",
"ipv4": "192.168.0.107"
},
"remoteEndpoint": {
"ipv4": "127.0.0.1",
"port": 49864
},
"tags": {
"http.method": "GET",
"http.path": "/order/create",
"mvc.controller.class": "OrderController",
"mvc.controller.method": "creatOrder"
},
"shared": true
},
{
"traceId": "94b954d843012ca9",
"parentId": "94b954d843012ca9",
"id": "bab70b1e69a5f3e3",
"kind": "CLIENT",
"name": "get",
"timestamp": 1569060494063693,
"duration": 10374,
"localEndpoint": {
"serviceName": "sleuth-search",
"ipv4": "192.168.0.107"
},
"tags": {
"http.method": "GET",
"http.path": "/cart/add/1"
}
},
{
"traceId": "94b954d843012ca9",
"parentId": "94b954d843012ca9",
"id": "90f7e5cfa89e0d80",
"kind": "CLIENT",
"name": "get",
"timestamp": 1569060494074966,
"duration": 2848,
"localEndpoint": {
"serviceName": "sleuth-search",
"ipv4": "192.168.0.107"
},
"tags": {
"http.method": "GET",
"http.path": "/order/create"
}
},
{
"traceId": "94b954d843012ca9",
"id": "94b954d843012ca9",
"kind": "SERVER",
"name": "get /search",
"timestamp": 1569060494062631,
"duration": 16332,
"localEndpoint": {
"serviceName": "sleuth-search",
"ipv4": "192.168.0.107"
},
"remoteEndpoint": {
"ipv6": "::1",
"port": 49859
},
"tags": {
"http.method": "GET",
"http.path": "/search",
"mvc.controller.class": "SearchController",
"mvc.controller.method": "search"
}
},
{
"traceId": "94b954d843012ca9",
"parentId": "bab70b1e69a5f3e3",
"id": "96387b33a823ca8f",
"kind": "CLIENT",
"name": "get",
"timestamp": 1569060494067090,
"duration": 3197,
"localEndpoint": {
"serviceName": "sleuth-cart",
"ipv4": "192.168.0.107"
},
"tags": {
"http.method": "GET",
"http.path": "/order/create"
}
},
{
"traceId": "94b954d843012ca9",
"parentId": "94b954d843012ca9",
"id": "bab70b1e69a5f3e3",
"kind": "SERVER",
"name": "get /cart/add/{cartid}",
"timestamp": 1569060494066140,
"duration": 8150,
"localEndpoint": {
"serviceName": "sleuth-cart",
"ipv4": "192.168.0.107"
},
"remoteEndpoint": {
"ipv4": "127.0.0.1",
"port": 49862
},
"tags": {
"http.method": "GET",
"http.path": "/cart/add/1",
"mvc.controller.class": "CartController",
"mvc.controller.method": "addToCart"
},
"shared": true
}
]
使用 RabbitMQ 情况
启动 zipkin,注意参数
java -jar zipkin-server-2.16.2-exec.jar --RABBIT_ADDRESSES=localhost:5672 --RABBIT_USER=guest --RABBIT_PASSWORD=guest --RABBIT_VIRTUAL_HOST=/
启动 rabbitmq
rabbitmq-server
在测试的时候发现 mq 和以上方式时延相差无几,但是随着线程数的增加也就是并发量的增加,mq 传输时延将会大大低于 http。