Flink限流使用GuavaFlinkConnectorRateLimiter

1.重写AbstractRichFunction中的open()方法，在处理数据前调用limiter.acquire(1);

private GuavaFlinkConnectorRateLimiter limiter;

@Override
public void open(Configuration parameters) throws Exception {
    limiter = new GuavaFlinkConnectorRateLimiter();
    limiter.setRate(3000);//表示task限流3000rec/s，不是subTask
    limiter.open(getRuntimeContext());
}
@Override
public void processElement(JSONObject value, Context ctx, Collector<Object> out) throws Exception{
    limiter.acquire(1);
    .........................

2.GuavaFlinkConnectorRateLimiter限流原理

 private RateLimiter rateLimiter;
 
 public void open(RuntimeContext runtimeContext) {
        this.runtimeContext = runtimeContext;
        localRateBytesPerSecond =
                globalRateBytesPerSecond / runtimeContext.getNumberOfParallelSubtasks();
        this.rateLimiter = RateLimiter.create(localRateBytesPerSecond);
}

调用limiter.open(getRuntimeContext())的源码，实际内部是RateLimiter,根据并行度算出subTask中RateLimiter的permitsPerSecond。