当我使用“ SlidingEventTimeWindows”时，Flink抛出“ java.lang.IllegalStateException：缓冲池被破坏”，但是当我更改为“ SlidingProcessingTimeWindows”时，一切正常。
以下是堆栈跟踪：
18:37:53,728 WARN org.apache.flink.streaming.api.operators.AbstractStreamOperator - Error while emitting latency marker.
java.lang.RuntimeException: Buffer pool is destroyed.
at org.apache.flink.streaming.runtime.io.RecordWriterOutput.emitLatencyMarker(RecordWriterOutput.java:147)
at org.apache.flink.streaming.api.operators.AbstractStreamOperator$CountingOutput.emitLatencyMarker(AbstractStreamOperator.java:683)
at org.apache.flink.streaming.api.operators.StreamSource$LatencyMarksEmitter$1.onProcessingTime(StreamSource.java:151)
at org.apache.flink.streaming.runtime.tasks.SystemProcessingTimeService$RepeatedTriggerTask.run(SystemProcessingTimeService.java:330)
at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:308)
at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:180)
at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:294)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:748)
Caused by: java.lang.IllegalStateException: Buffer pool is destroyed.
at org.apache.flink.runtime.io.network.buffer.LocalBufferPool.requestMemorySegment(LocalBufferPool.java:230)
at org.apache.flink.runtime.io.network.buffer.LocalBufferPool.requestBufferBuilderBlocking(LocalBufferPool.java:204)
at org.apache.flink.runtime.io.network.api.writer.RecordWriter.requestNewBufferBuilder(RecordWriter.java:213)
at org.apache.flink.runtime.io.network.api.writer.RecordWriter.sendToTarget(RecordWriter.java:144)
at org.apache.flink.runtime.io.network.api.writer.RecordWriter.randomEmit(RecordWriter.java:125)
at org.apache.flink.streaming.runtime.io.StreamRecordWriter.randomEmit(StreamRecordWriter.java:93)
at org.apache.flink.streaming.runtime.io.RecordWriterOutput.emitLatencyMarker(RecordWriterOutput.java:144)
... 10 more我终于解决了以下步骤。
首先，在My DataMockSource中将“ collect”替换为“ collectWithTimestamp”，用于生成流数据。执行此操作后，控制台中将消失“发出延迟标记时出错”。
其次，将BoundedOutOfOrdernessTimestampExtractor替换为用于EventTime处理的AscendingTimestampExtractor。在我的DataMockSource中，我生成数据并在同一时间发出，因此AscendingTimestampExtractor是生成水印的正确方法。
我将主要代码和完整项目发布在github上。希望对您有所帮助。
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime);
env.enableCheckpointing(10000); //
DataStreamSource<MockData> mockDataDataStreamSource = env.addSource(new DataMockSource());
mockDataDataStreamSource.assignTimestampsAndWatermarks(
new AscendingTimestampExtractor<MockData>() {
@Override
public long extractAscendingTimestamp(MockData element) {
return element.getTimestamp();
}
});
SingleOutputStreamOperator<Tuple2<String, Long>> countStream = mockDataDataStreamSource
.keyBy("country").window(
SlidingEventTimeWindows.of(Time.seconds(10), Time.seconds(10)))
// .allowedLateness(Time.seconds(5))
.process(
new FlinkEventTimeCountFunction()).name("count elements");
countStream.addSink(new SinkFunction<Tuple2<String, Long>>() {
@Override
public void invoke(Tuple2<String, Long> value, Context context) throws Exception {
System.out.println(value);
}
});
env.execute("count test ");我的DataMockSource在这里：
private volatile boolean running = true;
@Override
public void run(SourceContext sourceContext) throws Exception {
while (running){
MockData mockData = new MockData();
mockData.setAge(ThreadLocalRandom.current().nextInt(1,99));
mockData.setCountry("country "+ThreadLocalRandom.current().nextInt(2,5));
mockData.setId(ThreadLocalRandom.current().nextLong());
mockData.setTimestamp(Instant.now().toEpochMilli());
// emit record with timestamp
sourceContext.collectWithTimestamp(mockData,Instant.now().toEpochMilli());
// sourceContext.collect(mockData);
TimeUnit.SECONDS.sleep(3);
}
}
@Override
public void cancel() {
running = false;
}在事件时间中工作时，您需要安排在源中或使用assignTimestampsAndWatermarks进行时间戳提取和加水印。看来您没有这样做，这可以解释为什么您不会获得任何输出（永远不会触发事件时间窗口）。
另外，您的来源应有一个cancel方法。像这样：
private volatile boolean running = true;
@Override
public void run(SourceContext ctx) throws Exception {
while (running) {
...
}
}
@Override
public void cancel() {
running = false;
}
我认为这可以解释您所看到的异常。在作业开始自行关闭之后，源可能正在继续运行并发送延迟标记。
本文收集自互联网，转载请注明来源。
如有侵权，请联系 [email protected] 删除。
我来说两句