代码地址:https://github.com/zhaoyunxing92/spring-boot-learn-box/tree/master/spring-boot-sentinel/sentinel-apollo
使用apollo为sentinel做数据持久化,sentinel的使用可以看sentinel使用,本项目打算采用spring-cloud-alibaba跟sentinel整合
背景
我们的项目即将上线了,但是缺少一个流量防卫兵,这个很要命,于是一周前我就开始着手看sentinel
为什么是sentinel
正如你们所了解的市面上有很多服务降级方案,如我也用过的hystrix,但是它没有完善的持久方案和后台管理界面再加上我也看了一片sentinel作者宿何的一篇对比文章Sentinel-与-Hystrix-的对比,而且我们现在的技术不太适合使用spring cloud体系
参考文档
- sentinel动态规则扩展
- Sentinel使用Apollo存储规则
- spring-cloud-alibaba
整合流程
pom.xml
org.springframework.cloud
spring-cloud-starter-alibaba-sentinel
0.9.0.RELEASE
com.alibaba.csp
sentinel-datasource-apollo
1.6.0
application.yml
spring:
application:
name: spring-boot-sentinel-apollo
cloud:
sentinel:
transport:
port: 8719 # 向sentinel-dashboard传输数据的端口 默认:8719
dashboard: localhost:8100 # sentinel-dashboard
log:
dir: ./logs # 默认值${home}/logs/csp/
switch-pid: true # 日志带上线程id
datasource:
flow: # 流控规则
apollo:
namespaceName: application
flowRulesKey: flowRules
rule-type: flow #flow,degrade,authority,system, param-flow
degrade: # 熔断降级规则
apollo:
namespaceName: application
flowRulesKey: degrades
rule-type: degrade
authority: # 授权规则 未验证,官方不推荐
apollo:
namespaceName: application
flowRulesKey: authoritys
rule-type: authority
system: # 系统规则
apollo:
namespaceName: application
flowRulesKey: systems
rule-type: system
param-flow: # 热点规则
apollo:
namespaceName: application
flowRulesKey: paramflows
rule-type: param-flow
app:
id: ${spring.application.name}
apollo:
meta: http://127.0.0.1:8080
cacheDir: ./apolloconfig # 缓存文件位置
java
@SpringBootApplication
@EnableApolloConfig // 开启apollo
public class SpringSentinelApolloServer {
public static void main(String[] args) {
SpringApplication.run(SpringSentinelApolloServer.class, args);
}
}
jvm参数配置
-Denv=DEV
flow(流控规则)参数格式json
[
{
"resource": "/hello",
"limitApp": "default",
"grade": 1,
"count": 3,
"strategy": 0,
"controlBehavior": 0,
"clusterMode": false
}
]
flow(流控规则)参数规则说明
| 字段 | 描述 | 默认值 |
|---|---|---|
| resource | 资源名，即限流规则的作用对象 | |
| limitApp | 流控针对的调用来源，若为 default 则不区分调用来源 | default |
| grade | 限流阈值类型（QPS 或并发线程数）；0代表根据并发数量来限流，1代表根据QPS来进行流量控制 | QPS 模式 |
| count | 限流阈值 | |
| strategy | 调用关系限流策略 | |
| controlBehavior | 流量控制效果（直接拒绝、Warm Up、匀速排队） | 拒绝 |
| clusterMode | 是否为集群模式 |
system 系统规则参数格式,四个参数只能选择一个不能设置-1
[{"qps": 2}]
system参数列表
| 参数 | 说明 |
|---|---|
| avgLoad | 最大的 load |
| avgRt | 所有入口流量的平均响应时间 |
| maxThread | 入口流量的最大并发数 |
| qps | 所有入口资源的 QPS |
degrade 参数格式 json
[
{
"resource": "/rt",
"count": 50,
"timeWindow": 5,
"grade": 0
},
{
"resource": "/count",
"count": 5,
"timeWindow": 8,
"grade": 2
},
{
"resource": "/erro",
"count": 0.5,
"timeWindow": 5,
"grade": 1
}
]
degrade(熔断降级规则)参数规则说明
| 字段 | 描述 | 默认值 |
|---|---|---|
| resource | 资源名，即限流规则的作用对象 | |
| count | 阈值 | |
| grade | 降级模式，根据 RT (0)、异常数(2)、 异常比例(1) | RT (0) |
| timeWindow | 降级的时间，单位为 s |
param-flow(热点规则) json
[
{
"resource": "/hello",
"grade": 1,
"paramIdx": 1,
"count": 10,
"paramFlowItemList": []
}
]
param-flow(热点规则) 参数
| 字段 | 描述 | 默认值 |
|---|---|---|
| resource | 资源名，必填 | |
| grade | 限流模式 | qps(1) |
| paramIdx | 热点参数的索引，必填，对应 SphU.entry(xxx, args) 中的参数索引位置 |
|
| count | 限流阈值，必填 | |
| paramFlowItemList | 参数例外项，可以针对指定的参数值单独设置限流阈值，不受前面 count 阈值的限制。 |
apollo上配置
server.port = 7852
server.servlet.context-path = /sentinel
flowRules = [{"resource": "/hello","limitApp": "default","grade": 1,"count": 3,"strategy": 0,"controlBehavior": 0,"clusterMode": false}]
degrades = [{"resource": "/rt","count": 50,"timeWindow": 5,"grade": 0},{"resource": "/count","count": 5,"timeWindow": 8,"grade": 2},{"resource": "/erro","count": 0.5,"timeWindow": 5,"grade": 1}]
authoritys = [{"resource": "/hello","limitApp": "192.168.12.215","strategy": 1}]
paramflows = [{"resource": "/hello","grade": 1,"paramIdx": 1,"count": 10,"paramFlowItemList": []}]
systems = [{"qps": 20}]
拉去规则成功日志
2019-05-14 09:26:46.072 INFO 10100 --- [ main] o.s.c.a.s.c.SentinelDataSourceHandler : [Sentinel Starter] DataSource authority-sentinel-apollo-datasource load 1 AuthorityRule
2019-05-14 09:26:46.090 INFO 10100 --- [ main] o.s.c.a.s.c.SentinelDataSourceHandler : [Sentinel Starter] DataSource degrade-sentinel-apollo-datasource load 3 DegradeRule
2019-05-14 09:26:46.099 INFO 10100 --- [ main] o.s.c.a.s.c.SentinelDataSourceHandler : [Sentinel Starter] DataSource flow-sentinel-apollo-datasource load 1 FlowRule
2019-05-14 09:26:46.115 INFO 10100 --- [ main] o.s.c.a.s.c.SentinelDataSourceHandler : [Sentinel Starter] DataSource param-flow-sentinel-apollo-datasource load 1 ParamFlowRule
2019-05-14 09:26:46.122 INFO 10100 --- [ main] o.s.c.a.s.c.SentinelDataSourceHandler : [Sentinel Starter] DataSource system-sentinel-apollo-datasource load 1 SystemRule
测试接口
http://localhost:7852/sentinel/hello
最终效果图
sentinel-dashboard
apollo-dashboard
结尾
相信坚持看完的同学已经崩溃了,拉模式的配置太头疼了.的确为了完成这篇文章,里面的参数都是我在浏览器里面抓取过来的耗费了我大量的时间,巧的是sentinel官方也意识到了这点他们提供了推模式只不过需要修改sentinel-dashboard
的源码,后面我会把真理好的代码提交在sentinel-dashboard-apollo仓库,已经需改了一点,有兴趣的可以先去看