timeout
在调用第三方服务时有些情况需要对服务响应时间进行把控,当超时的情况下进行fallback的处理
下面来看下超时的案例
public class CommandTimeout extends HystrixCommand {
private final String name;
public CommandTimeout(String name) {
super(HystrixCommandGroupKey.Factory.asKey("ExampleGroup"));
this.name = name;
}
@Override
protected String run() {
System.out.println("aaaa");
try {
//sleep10秒强制超时,默认超时时间是1s
Thread.sleep(10000);
} catch (InterruptedException e) {
e.printStackTrace();
}
System.out.println("run end");
return "";
}
@Override
protected String getFallback() {
return "Hello Failure " + name + "!";
}
}
接下来是测试方法
@Test
public void testSynchronous() throws InterruptedException {
System.out.println(new CommandHelloWorld("World").execute());
}
java.lang.InterruptedException: sleep interrupted
at java.lang.Thread.sleep(Native Method)
at hystrix.CommandHelloWorld.run(CommandHelloWorld.java:32)
at hystrix.CommandHelloWorld.run(CommandHelloWorld.java:1)
...
run end
Hello Failure World!
可以看到sleep被强制interrupted,并且调用的输出也变成了fallback方法的返回值
如何查看是哪里调用的interrupt方法
这里顺便说下如何看是哪个方法调用的interrupt
根据stackoverflow的一个答案,没有直接的方法来断点到interrupt的方法,只能通过在Thread的interrupt方法上打断点,再反向看栈信息得知哪里中断当前线程。
如何改变timeout设置
在HystrixCommand的够着方法中可以在第二个参数配置一个timeout的毫秒数
public CommandHelloWorld(String name) {
super(HystrixCommandGroupKey.Factory.asKey("ExampleGroup"),1000000000);
this.name = name;
}
这个构造方法是在调用AbastractCommand的构造方法时将毫秒数配置在CommandProperties中,如下:
super(group, null, null, null, null, HystrixCommandProperties.Setter().withExecutionTimeoutInMilliseconds(executionIsolationThreadTimeoutInMilliseconds), null, null, null, null, null, null);
Hystrix的timeout是怎么运行的
在运行对应的command时,Hystrix会通过HystrixObservableTimeoutOperator注册一个Timer到一个定时线程池中,当超时后会启用一个HystrixTimer线程来interruptCommand的执行
//创建一个TimerListener
public Reference addTimerListener(final TimerListener listener) {
startThreadIfNeeded();
// add the listener
Runnable r = new Runnable() {
@Override
public void run() {
try {
listener.tick();
} catch (Exception e) {
logger.error("Failed while ticking TimerListener", e);
}
}
};
//通过ScheduledThreadPoolExecutor在到达超时时间时运行上面的listener.tick,而时间是从listener的getIntervalTimeInMilliseconds方法中获得的
ScheduledFuture> f = executor.get().getThreadPool().scheduleAtFixedRate(r, listener.getIntervalTimeInMilliseconds(), listener.getIntervalTimeInMilliseconds(), TimeUnit.MILLISECONDS);
//返回包含timer的Reference,在任务在规定时间内完成是用于cancel超时处理
return new TimerReference(listener, f);
}
//下面是上面listener的定义
TimerListener listener = new TimerListener() {
@Override
public void tick() {
// if we can go from NOT_EXECUTED to TIMED_OUT then we do the timeout codepath
// otherwise it means we lost a race and the run() execution completed or did not start
//这里使用CAS的操作将将状态设置为TIME_OUT,使用CAS的原因是如果运行成功而timeout没有被取消时不会标记任务超时
if (originalCommand.isCommandTimedOut.compareAndSet(TimedOutStatus.NOT_EXECUTED, TimedOutStatus.TIMED_OUT)) {
// report timeout failure
originalCommand.eventNotifier.markEvent(HystrixEventType.TIMEOUT, originalCommand.commandKey);
// shut down the original request
//内部会取消当前并调用fallback
s.unsubscribe();
timeoutRunnable.run();
//if it did not start, then we need to mark a command start for concurrency metrics, and then issue the timeout
}
}
@Override
public int getIntervalTimeInMilliseconds() {
//这里从command的配置中获得配置的超时时间
return originalCommand.properties.executionTimeoutInMilliseconds().get();
}
};
上面是Command超时后的处理操作,当Command在时间内完成时会调用TimeReference的clear方法,内部调用了future的cancel来取消timer的超时任务
private static class TimerReference extends SoftReference {
private final ScheduledFuture> f;
//保存scheduledFuture
TimerReference(TimerListener referent, ScheduledFuture> f) {
super(referent);
this.f = f;
}
@Override
public void clear() {
super.clear();
// stop this ScheduledFuture from any further executions
//停止scheduledFuture
f.cancel(false);
}
}