文章目录
一、目的概述
为了验证Ribbon客户端负载均衡策略在负载节点失效
的情况下,是否具有故障转移
的功能,进行了以下代码验证!
二、验证步骤
1、源码下载
bash
git clone https://gitee.com/00fly/microservice-all-in-one.git
https://gitee.com/00fly/microservice-all-in-one/tree/master/ribbon-demo-simple
2、导入IDE
3、运行前修改配置
根据调用关系,我们需要启动2个user服务,为了方便调试我们这边分别启动8081、8082端口的user服务,并在movie模块中,设置负载节点地址为:127.0.0.1:8081,127.0.0.1:8082
微服务movie 微服务user 微服务user
以eclipse
为例简要说明
查看环境配置
打开Dashboard,选择Duplicate config
选择open Config
选择Profile设置为dev
全部启动
docker部署 相对简单,编排文件为
https://gitee.com/00fly/microservice-all-in-one/blob/master/ribbon-demo-simple/docker/docker-compose.yml
yaml
version: '3.8'
services:
#负载均衡节点
ribbon-user-simple-0:
image: registry.cn-shanghai.aliyuncs.com/00fly/ribbon-user-simple:0.0.1
container_name: ribbon-user-simple-0
deploy:
resources:
limits:
cpus: '1'
memory: 200M
reservations:
memory: 180M
restart: on-failure
logging:
driver: json-file
options:
max-size: 5m
max-file: '1'
#负载均衡节点
ribbon-user-simple-1:
image: registry.cn-shanghai.aliyuncs.com/00fly/ribbon-user-simple:0.0.1
container_name: ribbon-user-simple-1
deploy:
resources:
limits:
cpus: '1'
memory: 200M
reservations:
memory: 180M
restart: on-failure
logging:
driver: json-file
options:
max-size: 5m
max-file: '1'
#调用方
ribbon-movie-simple:
image: registry.cn-shanghai.aliyuncs.com/00fly/ribbon-movie-simple:0.0.1
container_name: ribbon-movie-simple
deploy:
resources:
limits:
cpus: '1'
memory: 200M
reservations:
memory: 180M
ports:
- 8090:8082
environment:
USER_SERVERS: ribbon-user-simple-0:8081,ribbon-user-simple-1:8081
restart: on-failure
logging:
driver: json-file
options:
max-size: 5m
max-file: '1'
4、策略说明
- RandomRule 实现从服务实例清单中随机选择一个服务实例的功能。
- RoundRobinRule 实现了按照线性轮询的方式依次选择每个服务实例的功能。
- RetryRule 实现了一个具备重试机制的实例选择功能。
- WeightedResponseTimeRule是对 RoundRobinRule 的拓展,增加了根据实例的运行情况来计算权重,并根据权重来挑选实例。
- ClientConfigEnableRoundRobinRule 通过继承该策略,在子类中做一些高级策略时有可能会存在一些无法实施的情况,那么就可以用父类的实现作为备选(线性轮询机制)。
- BestAvailableRule 通过遍历负载均衡器中维护的所有服务实例,会过滤掉故障的实例,并找出并发请求数最小的一个,所以该策略的特性是可选出最空闲的实例。
- PredicateBasedRule 先通过子类实现中的 Predicate 逻辑来过滤一部分服务实例,然后再以线性轮询的方式从过滤后的实例清单中选出一个。
- AvailabilityFilteringRule 通过线性抽样的方式直接尝试寻找可用且较空闲的实例来使用。
- ZoneAvoidanceRule 根据负载情况选择可用区
5、修改策略
修改这边的负载均衡策略
打开页面
停止8081或8082端口服务,重新调试,返回结果如下:
三、最终结论
RandomRule、RoundRobinRule 策略不具备故障转移能力
RetryRule、WeightedResponseTimeRule等虽然具有故障转移,但是故障转移的时间太长,并且故障恢复后,重新选中该恢复的节点所需时间也较长。
各种策略的表现。大家可以自行研究测试。
四、改进措施
1. 思路分析
采用多线程,多个节点同时检测,返回最快响应的节点
采用多线程,定义超时时间,返回超时时间之内有响应的节点, 后续根据规则选择1个节点
2. 核心代码
NodeController.java
java
import java.nio.charset.StandardCharsets;
import java.util.ArrayList;
import java.util.List;
import java.util.concurrent.Callable;
import java.util.concurrent.CopyOnWriteArrayList;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.Future;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.TimeoutException;
import java.util.stream.Collectors;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.http.MediaType;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.PathVariable;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RestController;
import org.springframework.web.reactive.function.client.WebClient;
import com.itmuch.cloud.study.user.entity.User;
import io.swagger.annotations.Api;
import io.swagger.annotations.ApiOperation;
import lombok.extern.slf4j.Slf4j;
import reactor.core.publisher.Mono;
@Slf4j
@Api(tags = "负载均衡节点")
@RestController
@RequestMapping("/node")
public class NodeController
{
@Autowired
private WebClient webClient;
@Value("${microservice-ribbon-user.ribbon.listOfServers}")
private List<String> listOfServers;
private ExecutorService executorService = Executors.newFixedThreadPool(10);
@ApiOperation("查询用户")
@GetMapping("/user/{id}")
public List<User> findById(@PathVariable Long id)
throws InterruptedException
{
// WebClient支持异步
List<User> users = new CopyOnWriteArrayList<User>();
listOfServers.stream()
.forEach(hostWithPort -> webClient.get()
.uri(String.format("http://%s/%s", hostWithPort, id))// URI
.acceptCharset(StandardCharsets.UTF_8)
.accept(MediaType.APPLICATION_JSON)
.retrieve()
.bodyToMono(User.class)
.subscribe(resp -> users.add(resp)));
int index = 0;
while (users.isEmpty() && (index++) < 100)
{
TimeUnit.MILLISECONDS.sleep(10);
log.info("index:{}, waitting......", index);
}
if (users.isEmpty())
{
throw new RuntimeException("查询超时,无返回值");
}
return users;
}
@ApiOperation("查询用户 by execute")
@GetMapping("/v0/user/{id}")
public List<User> findByExecute(@PathVariable Long id)
throws InterruptedException
{
// List<User> users = new ArrayList<User>();
// TODO ArrayList users一定概率有null值
// 原因:通过new ArrayList<>()初始化的大小是0,首次插入触发扩容,并发可能导致出现null值
List<User> users = new CopyOnWriteArrayList<User>();
listOfServers.stream()
.forEach(hostWithPort -> executorService.execute(() -> webClient.get()
.uri(String.format("http://%s/%s", hostWithPort, id))// URI
.acceptCharset(StandardCharsets.UTF_8)
.accept(MediaType.APPLICATION_JSON)
.retrieve()
.bodyToMono(User.class)
.subscribe(resp -> users.add(resp))));
int index = 0;
while (users.isEmpty() && (index++) < 100)
{
TimeUnit.MILLISECONDS.sleep(10);
log.info("index:{}, waitting......", index);
}
if (users.isEmpty())
{
throw new RuntimeException("查询超时,无返回值");
}
return users;
}
@ApiOperation("查询用户 by submit")
@GetMapping("/v1/user/{id}")
public List<User> findBySubmit(@PathVariable Long id)
throws InterruptedException
{
List<User> users = new CopyOnWriteArrayList<User>();
listOfServers.stream()
.forEach(hostWithPort -> executorService.submit(() -> webClient.get()
.uri(String.format("http://%s/%s", hostWithPort, id))// URI
.acceptCharset(StandardCharsets.UTF_8)
.accept(MediaType.APPLICATION_JSON)
.retrieve()
.bodyToMono(User.class)
.subscribe(resp -> users.add(resp)), users));
int index = 0;
while (users.isEmpty() && (index++) < 100)
{
TimeUnit.MILLISECONDS.sleep(10);
log.info("index:{}, waitting......", index);
}
if (users.isEmpty())
{
throw new RuntimeException("查询超时,无返回值");
}
return users;
}
@ApiOperation("查询用户 by invokeAny")
@GetMapping("/v2/user/{id}")
public User findByInvokeAny(@PathVariable Long id)
throws InterruptedException, ExecutionException, TimeoutException
{
return executorService.invokeAny(listOfServers.stream().map(hostWithPort -> new Callable<User>()
{
@Override
public User call()
{
Mono<User> mono = webClient.get()
.uri(String.format("http://%s/%s", hostWithPort, id))// URI
.acceptCharset(StandardCharsets.UTF_8)
.accept(MediaType.APPLICATION_JSON)
.retrieve()
.bodyToMono(User.class);
return mono.block();
}
}).collect(Collectors.toList()), 1000, TimeUnit.MILLISECONDS);
}
@ApiOperation("查询用户 by invokeAll")
@GetMapping("/v3/user/{id}")
public List<User> findByInvokeAll(@PathVariable Long id)
throws InterruptedException
{
List<Future<User>> futures = executorService.invokeAll(listOfServers.stream().map(hostWithPort -> new Callable<User>()
{
@Override
public User call()
{
Mono<User> mono = webClient.get()
.uri(String.format("http://%s/%s", hostWithPort, id))// URI
.acceptCharset(StandardCharsets.UTF_8)
.accept(MediaType.APPLICATION_JSON)
.retrieve()
.bodyToMono(User.class);
return mono.block();
}
}).collect(Collectors.toList()), 1000, TimeUnit.MILLISECONDS);
List<User> users = new ArrayList<User>();
for (Future<User> future : futures)
{
try
{
users.add(future.get());
}
catch (Exception e)
{
log.error(e.getMessage(), e);
}
}
return users;
}
}
3. 测试页面
有任何问题和建议,都可以向我提问讨论,大家一起进步,谢谢!
-over-