本文属于sentinel学习笔记系列。网上看到吴就业老师的专栏,写的好值得推荐,我整理的有所删减,推荐看原文。
https://blog.csdn.net/baidu_28523317/category_10400605.html
sentinel 实现限流降级、熔断降级、黑白名单限流降级、系统自适应限流降级以及热点参数限流降级都是由 ProcessorSlot、Checker、Rule、RuleManager 组合完成。ProcessorSlot 作为调用链路的切入点,负责调用 Checker 检查当前请求是否可以放行;Checker 则根据资源名称从 RuleManager 中拿到为该资源配置的 Rule(规则),取 ClusterNode 统计的实时指标数据与规则对比,如果达到规则的阈值则抛出 Block 异常,抛出 Block 异常意味着请求被拒绝,也就实现了限流或熔断。
可以总结为以下三个步骤:
- 在 ProcessorSlot#entry 方法中调用 Checker#check 方法,并将 DefaultNode 传递给 Checker。
- Checker 从 DefaultNode 拿到 ClusterNode,并根据资源名称从 RuleManager 获取为该资源配置的规则。
- Checker 从 ClusterNode 中获取当前时间窗口的某项指标数据(QPS、avgRt 等)与规则的阈值对比,如果达到规则的阈值则抛出 Block 异常(也有可能将 check 交给 Rule 去实现)。
限流规则与规则配置加载器
rule
规则是围绕资源配置的,接口Rule 只定义获取资源。
public interface Rule {/*** Get target resource of this rule.** @return target resource of this rule*/String getResource();}
public abstract class AbstractRule implements Rule {/*** rule id. */private Long id;/*** Resource name. 资源名*/private String resource;/*** 流控对应的调用来源*/private String limitApp;
Rule、AbstractRule 与其它实现类的关系如下图所示:
FlowRule 是限流规则配置类,FlowRule 继承 AbstractRule 并实现 Rule 接口。FlowRule 源码如下
public class FlowRule extends AbstractRule {public FlowRule() {super();setLimitApp(RuleConstant.LIMIT_APP_DEFAULT);}public FlowRule(String resourceName) {super();setResource(resourceName);setLimitApp(RuleConstant.LIMIT_APP_DEFAULT);}/*** The threshold type of flow control (0: thread count, 1: QPS).* 限流阈值类型*/private int grade = RuleConstant.FLOW_GRADE_QPS;/*** Flow control threshold count. 限流阈值 配置的是qps类型则代表qps的值;配置的是线程数类型则代表线程数*/private double count;/*** Flow control strategy based on invocation chain.* 流控限流策略* {@link RuleConstant#STRATEGY_DIRECT} for direct flow control (by origin);* {@link RuleConstant#STRATEGY_RELATE} for relevant flow control (with relevant resource);* {@link RuleConstant#STRATEGY_CHAIN} for chain flow control (by entrance resource).*/private int strategy = RuleConstant.STRATEGY_DIRECT;/*** Reference resource in flow control with relevant resource or context.* 关联流控的资源*/private String refResource;/*** Rate limiter control behavior. 流控效果控制* 0. default(reject directly), 1. warm up, 2. rate limiter, 3. warm up + rate limiter*/private int controlBehavior = RuleConstant.CONTROL_BEHAVIOR_DEFAULT;//对应流控效果为Warm Up情况下,冷启动时长(预热时长),单位秒private int warmUpPeriodSec = 10;/*** Max queueing time in rate limiter behavior.* 对应流控效果为排队等待情况下,出现的超时时间*/private int maxQueueingTimeMs = 500;// 对应新增流控规则页面的是否集群private boolean clusterMode;/*** Flow rule config for cluster mode.集群流控的相关配置*/private ClusterFlowConfig clusterConfig;/*** The traffic shaping (throttling) controller. 流量整形的实现,不同流控效果有不同算法*/private TrafficShapingController controller;
字段属性有些多,可以对比sentinel 限流保护-笔记-CSDN博客 跟官网文档来理解
RuleManager
Sentinel 中用来管理规则配置的类都以规则类的名称+Manger 命名,用来加载限流规则配置以及缓存限流规则配置的类为 FlowRuleManager,其部分源码如下:
public class FlowRuleManager {// 缓存限流规则private static volatile Map<String, List<FlowRule>> flowRules = new HashMap<>();// PropertyListener 监听器private static final FlowPropertyListener LISTENER = new FlowPropertyListener();//SentinelProperty ,默认的 DynamicSentinelPropertyprivate static SentinelProperty<List<FlowRule>> currentProperty = new DynamicSentinelProperty<List<FlowRule>>();/** the corePool size of SCHEDULER must be set at 1, so the two task ({@link #startMetricTimerListener()} can run orderly by the SCHEDULER **/@SuppressWarnings("PMD.ThreadPoolCreationRule")private static final ScheduledExecutorService SCHEDULER = Executors.newScheduledThreadPool(1,new NamedThreadFactory("sentinel-metrics-record-task", true));static {//给默认的 SentinelProperty 注册监听器(FlowPropertyListener)currentProperty.addListener(LISTENER);startMetricTimerListener();}。。。public static List<FlowRule> getRules() {List<FlowRule> rules = new ArrayList<FlowRule>();for (Map.Entry<String, List<FlowRule>> entry : flowRules.entrySet()) {rules.addAll(entry.getValue());}return rules;}//更新规则public static void loadRules(List<FlowRule> rules) {currentProperty.updateValue(rules);}
我们之前demo对用FlowRuleManager.loadRules()来更新规则生效,注意1.8.6版本这里面
SentinelProperty ,SentinelProperty 是 Sentinel 提供的一个接口,可注册到 Sentinel 提供的各种规则的 Manager,例如 FlowRuleManager,并且可以给 SentinelProperty 添加监听器,在配置改变时,你可以调用 SentinelProperty#updateValue 方法,由它负责调用监听器去更新规则,而不需要调用 FlowRuleManager#loadRules 方法。这块先先不展开梳理,后面结合nacos再看。
限流处理器插槽:FlowSlot
FlowSlot 是实现限流功能的切入点,它作为 ProcessorSlot 插入到 ProcessorSlotChain 链表中,在 entry 方法中调用 Checker 去判断是否需要拒绝当前请求,如果需要拒绝请求则抛出 Block 异常。FlowSlot 的源码如下:
@Spi(order = Constants.ORDER_FLOW_SLOT)
public class FlowSlot extends AbstractLinkedProcessorSlot<DefaultNode> {private final FlowRuleChecker checker;public FlowSlot() {this(new FlowRuleChecker());}/*** Package-private for test.** @param checker flow rule checker* @since 1.6.1*/FlowSlot(FlowRuleChecker checker) {AssertUtil.notNull(checker, "flow checker should not be null");this.checker = checker;}@Overridepublic void entry(Context context, ResourceWrapper resourceWrapper, DefaultNode node, int count,boolean prioritized, Object... args) throws Throwable {checkFlow(resourceWrapper, context, node, count, prioritized);fireEntry(context, resourceWrapper, node, count, prioritized, args);}//校验是否限流void checkFlow(ResourceWrapper resource, Context context, DefaultNode node, int count, boolean prioritized)throws BlockException {checker.checkFlow(ruleProvider, resource, context, node, count, prioritized);}@Overridepublic void exit(Context context, ResourceWrapper resourceWrapper, int count, Object... args) {fireExit(context, resourceWrapper, count, args);}// 规则生产者private final Function<String, Collection<FlowRule>> ruleProvider = new Function<String, Collection<FlowRule>>() {@Override // 参数为资源名称public Collection<FlowRule> apply(String resource) {// Flow rule map should not be null.Map<String, List<FlowRule>> flowRules = FlowRuleManager.getFlowRuleMap();return flowRules.get(resource);}};
}
限流规则检查器:FlowRuleChecker
FlowRuleChecker 负责判断是否需要拒绝当前请求,方法很多,先看看调用checkFlow
public class FlowRuleChecker {public void checkFlow(Function<String, Collection<FlowRule>> ruleProvider, ResourceWrapper resource,Context context, DefaultNode node, int count, boolean prioritized) throws BlockException {if (ruleProvider == null || resource == null) {return;}// 获取匹配的规则Collection<FlowRule> rules = ruleProvider.apply(resource.getName());if (rules != null) {//遍历规则for (FlowRule rule : rules) { // 检查规则能否通过if (!canPassCheck(rule, context, node, count, prioritized)) {throw new FlowException(rule.getLimitApp(), rule);}}}}
注意:遍历限流规则,只要有一个限流规则达到限流阈值即可抛出 FlowException,使用 FlowException 目的是标志当前请求因为达到限流阈值被拒绝,FlowException 是 BlockException 的子类;
canPassCheck 方法返回 true 说明允许请求通过,反之则不允许通过。canPassCheck 方法源码如下:
public boolean canPassCheck(/*@NonNull*/ FlowRule rule, Context context, DefaultNode node, int acquireCount,boolean prioritized) {String limitApp = rule.getLimitApp();//当前限流规则只对哪个调用来源生效,一般不为null,默认为“default”(不限定调用来源)if (limitApp == null) {return true;}// 集群模式下的规则检测if (rule.isClusterMode()) {return passClusterCheck(rule, context, node, acquireCount, prioritized);}//单机模式下规则检测return passLocalCheck(rule, context, node, acquireCount, prioritized);}
先不讨论集群限流的情况,看看单机的passLocalCheck
private static boolean passLocalCheck(FlowRule rule, Context context, DefaultNode node, int acquireCount,boolean prioritized) {//根据调用来源和“调用关系限流策略”选择 DefaultNode;Node selectedNode = selectNodeByRequesterAndStrategy(rule, context, node);if (selectedNode == null) {return true;}return rule.getRater().canPass(selectedNode, acquireCount, prioritized);}
- selectNodeByRequesterAndStrategy返回ClusterBuilderSlot阶段生成的ClusterNode
-
getRater 返回TrafficShapingController,在默认模式下返回流控效果策略DefaultController。DefaultController#canPass 完成canPassCheck。
下面分别看看这两个方法。
流控节点选择
selectNodeByRequesterAndStrategy 方法的实现有多种情况。原码如下:
public class FlowRuleChecker {static Node selectNodeByRequesterAndStrategy(/*@NonNull*/ FlowRule rule, Context context, DefaultNode node) {// The limit app should not be empty.限流规则针对哪个来源生效String limitApp = rule.getLimitApp();// 基于调用关系的限流策略int strategy = rule.getStrategy();// 远程来源String origin = context.getOrigin();if (limitApp.equals(origin) && filterOrigin(origin)) {if (strategy == RuleConstant.STRATEGY_DIRECT) {// 1 Matches limit origin, return origin statistic node.return context.getOriginNode();}//2return selectReferenceNode(rule, context, node);} else if (RuleConstant.LIMIT_APP_DEFAULT.equals(limitApp)) {if (strategy == RuleConstant.STRATEGY_DIRECT) {//3 Return the cluster node.return node.getClusterNode();}//4return selectReferenceNode(rule, context, node);} else if (RuleConstant.LIMIT_APP_OTHER.equals(limitApp)&& FlowRuleManager.isOtherOrigin(origin, rule.getResource())) {if (strategy == RuleConstant.STRATEGY_DIRECT) {// 5return context.getOriginNode();}//6return selectReferenceNode(rule, context, node);}return null;}static Node selectReferenceNode(FlowRule rule, Context context, DefaultNode node) {String refResource = rule.getRefResource();int strategy = rule.getStrategy();if (StringUtil.isEmpty(refResource)) {return null;}if (strategy == RuleConstant.STRATEGY_RELATE) {return ClusterBuilderSlot.getClusterNode(refResource);}if (strategy == RuleConstant.STRATEGY_CHAIN) {if (!refResource.equals(context.getName())) {return null;}return node;}// No node.return null;}
如果当前限流规则的 limitApp 为 default,则说明该限流规则对任何调用来源都生效,针对所有调用来源限流,否则只针对指定调用来源限流。
1 如果调用来源与当前限流规则的 limitApp 相等,且 strategy 为 STRATEGY_DIRECT,则使用调用来源的 StatisticNode,实现针对调用来源限流。
2 前置条件与(1)相同,依然是针对来源限流。selectReferenceNode
- strategy 为 STRATEGY_RELATE:使用引用资源的 ClusterNode;
- strategy 为 STRATEGY_CHAIN:使用当前资源的 DefauleNode。
3当 limitApp 为 default 时,针对所有来源限流。如果 strategy 为 STRATEGY_DIRECT,则使用当前资源的 ClusterNode。
4 前置条件与(3)相同,依然是针对所有来源限流。selectReferenceNode
5 如果 limitApp 为 other,且该资源的所有限流规则都没有针对当前的调用来源限流。如果 strategy 为 STRATEGY_DIRECT,则使用 origin 的 StatisticNode。
6 前置条件与(5)一样。selectReferenceNode
从 selectNodeByRequesterAndStrategy 方法可以看出,Sentinel 之所以针对每个资源统计访问来源的指标数据,也是为了实现对丰富的限流策略的支持.比如针对调用来源限流可限制并发量较高的来源服务的请求,而对并发量低的来源服务的请求可不限流,或者是对一些并没有那么重要的来源服务限流。
TrafficShapingController
Sentinel 支持对超出限流阈值的流量采取效果控制器控制这些流量,流量效果控制支持:直接拒绝、Warm Up(冷启动)、匀速排队。对应 FlowRule 中的 controlBehavior 字段。在调用 FlowRuleManager#loadRules 方法时,FlowRuleManager 会将限流规则配置的 controlBehavior 转为对应的 TrafficShapingController。
public interface TrafficShapingController {/*** Check whether given resource entry can pass with provided count.* 判断当前请求是否能通过* @param node resource node* @param acquireCount count to acquire* @param prioritized whether the request is prioritized* @return true if the resource entry can pass; false if it should be blocked*/boolean canPass(Node node, int acquireCount, boolean prioritized);/*** Check whether given resource entry can pass with provided count.* 判断当前请求是否能通过* @param node resource node* @param acquireCount count to acquire* @return true if the resource entry can pass; false if it should be blocked*/boolean canPass(Node node, int acquireCount);
}
DefaultController
DefaultController 是默认使用的流量效果控制器,直接拒绝超出阈值的请求。当 QPS 超过限流规则配置的阈值,新的请求就会被立即拒绝,抛出 FlowException。
@Overridepublic boolean canPass(Node node, int acquireCount, boolean prioritized) {// 获取当前已使用的token:qps 算每秒被放行的请求数,threads 统计的当前并行占用的线程数int curCount = avgUsedTokens(node);//当前已使用token + 获取的token 大于token数量的场景if (curCount + acquireCount > count) {//qps 且prioritized 参数的值为 true(有优先级的请求可以占用未来时间窗口的统计指标)if (prioritized && grade == RuleConstant.FLOW_GRADE_QPS) {long currentTime;long waitInMs;currentTime = TimeUtil.currentTimeMillis();//当前请求需要等待的时间,单位毫秒waitInMs = node.tryOccupyNext(currentTime, acquireCount, count);if (waitInMs < OccupyTimeoutProperty.getOccupyTimeout()) {//将休眠之后对应的时间窗口的 pass(通过)这项指标数据的值加上 acquireCountnode.addWaitingRequest(currentTime + waitInMs, acquireCount);// 添加占用未来的 pass 指标的数量node.addOccupiedPass(acquireCount);// 休眠等待,当前线程阻塞sleep(waitInMs);// PriorityWaitException indicates that the request will pass after waiting for {@link @waitInMs}.//休眠结束后抛出 PriorityWait 异常,表示当前请求是等待了 waitInMs 之后通过的throw new PriorityWaitException(waitInMs);}}return false;}return true;}private int avgUsedTokens(Node node) {if (node == null) {return DEFAULT_AVG_USED_TOKENS;}return grade == RuleConstant.FLOW_GRADE_THREAD ? node.curThreadNum() : (int)(node.passQps());}
一般情况下,prioritized 参数的值为 false,所以这个 canPass 方法实现的流量效果就是直接拒绝。
限于篇幅,其他限流RateLimiterController、WarmUpController、WarmUpRateLimiterController 待整理。