flink中的流式实现

时间:2016-04-16 16:39:08

标签: apache-flink flink-streaming

我的任务是从本地驱动器读取csv文件。它包含以特定间隔定期更新的库存数据。接下来,对读取的数据执行一些操作,并定期写回新的csv文件。

我的代码改编自 WikipediaAnalysis 代码,问题是执行只发生一次。在初始运行之后,程序结束并且不会保持运行状态。它不会像WikipediaAnalysis那样定期运行5秒(我已设置)。以下是我的代码:

StreamExecutionEnvironment see = StreamExecutionEnvironment.getExecutionEnvironment();
DataStream<String> dataStream = getCsvDataStream(see);
DataStream<Stock> edits = dataStream.flatMap(new Tokenizer());
@SuppressWarnings("serial")
KeyedStream<Stock, String> keyedEdits = edits.keyBy(new KeySelector<Stock, String>() {
            @Override
            public String getKey(Stock event) {
                java.util.Date time = new java.util.Date((long) event.getTimeStamp() * 1000);
                return time.toString();
            }
        });

@SuppressWarnings("serial")
DataStream<Tuple2<Double, Double>> result = keyedEdits.timeWindow(Time.seconds(5))
                .fold(new Tuple2<>(0.0d, 0.0d), new FoldFunction<Stock, Tuple2<Double, Double>>() {
                    @Override
                    public Tuple2<Double, Double> fold(Tuple2<Double, Double> acc, Stock event) {
                        acc.f0 = event.getClose();
                        acc.f1 = event.getOpen();                           
                        return acc;
                    }
                });

你能建议!!

0 个答案:

没有答案