本文主要研究一下PowerJob的UseCacheLock
tech/powerjob/server/core/lock/UseCacheLock.java
@Target(ElementType.METHOD)
@Retention(RetentionPolicy.RUNTIME)
public @interface UseCacheLock {
String type();
String key();
int concurrencyLevel();
}
UseCacheLock注解定义了type、key、concurrencyLevel属性
tech/powerjob/server/core/lock/UseCacheLockAspect.java
@Slf4j
@Aspect
@Component
@Order(1)
@RequiredArgsConstructor
public class UseCacheLockAspect {
private final MonitorService monitorService;
private final Map<String, Cache<String, ReentrantLock>> lockContainer = Maps.newConcurrentMap();
private static final long SLOW_THRESHOLD = 100;
@Around(value = "@annotation(useCacheLock))")
public Object execute(ProceedingJoinPoint point, UseCacheLock useCacheLock) throws Throwable {
Cache<String, ReentrantLock> lockCache = lockContainer.computeIfAbsent(useCacheLock.type(), ignore -> {
int concurrencyLevel = useCacheLock.concurrencyLevel();
log.info("[UseSegmentLockAspect] create Lock Cache for [{}] with concurrencyLevel: {}", useCacheLock.type(), concurrencyLevel);
return CacheBuilder.newBuilder()
.initialCapacity(300000)
.maximumSize(500000)
.concurrencyLevel(concurrencyLevel)
.expireAfterWrite(30, TimeUnit.MINUTES)
.build();
});
final Method method = AOPUtils.parseMethod(point);
Long key = AOPUtils.parseSpEl(method, point.getArgs(), useCacheLock.key(), Long.class, 1L);
final ReentrantLock reentrantLock = lockCache.get(String.valueOf(key), ReentrantLock::new);
long start = System.currentTimeMillis();
reentrantLock.lockInterruptibly();
try {
long timeCost = System.currentTimeMillis() - start;
if (timeCost > SLOW_THRESHOLD) {
final SlowLockEvent slowLockEvent = new SlowLockEvent()
.setType(SlowLockEvent.Type.LOCAL)
.setLockType(useCacheLock.type())
.setLockKey(String.valueOf(key))
.setCallerService(method.getDeclaringClass().getSimpleName())
.setCallerMethod(method.getName())
.setCost(timeCost);
monitorService.monitor(slowLockEvent);
log.warn("[UseSegmentLockAspect] wait lock for method({}#{}) cost {} ms! key = '{}', args = {}, ", method.getDeclaringClass().getSimpleName(), method.getName(), timeCost,
key,
JSON.toJSONString(point.getArgs()));
}
return point.proceed();
} finally {
reentrantLock.unlock();
}
}
}
UseCacheLockAspect拦截@UseCacheLock注解,它定义了lockContainer维护了useCacheLock.type()与Cache的关系;Cache采用的guava的Cache,其initialCapacity为300000,maximumSize为500000,expireAfterWrite为30分钟;Cache的key为lock key,value为ReentrantLock;其execute方法主要是先执行reentrantLock.lockInterruptibly(),然后执行point.proceed(),最后reentrantLock.unlock();执行point.proceed()之前还判断了一下加锁耗时,若超过SLOW_THRESHOLD(
100ms
)则通过monitorService.monitor上报SlowLockEvent
@UseCacheLock(type = "processJobInstance", key = "#instanceId", concurrencyLevel = 1024)
public void redispatchAsync(Long instanceId, int originStatus) {
// 将状态重置为等待派发
instanceInfoRepository.updateStatusAndGmtModifiedByInstanceIdAndOriginStatus(instanceId, originStatus, InstanceStatus.WAITING_DISPATCH.getV(), new Date());
}
key支持SpEl
PowerJob的UseCacheLock注解定义了type、key、concurrencyLevel属性;UseCacheLockAspect拦截@UseCacheLock注解,它定义了lockContainer维护了useCacheLock.type()与Cache的关系;而Cache的key为lock key,value为ReentrantLock,最后是通过reentrantLock.lockInterruptibly()加锁。
原创声明:本文系作者授权腾讯云开发者社区发表,未经许可,不得转载。
如有侵权,请联系 cloudcommunity@tencent.com 删除。
原创声明:本文系作者授权腾讯云开发者社区发表,未经许可,不得转载。
如有侵权,请联系 cloudcommunity@tencent.com 删除。