在 Apache Flink Java 中創建翻轉 window 時出錯

Question

我嘗試在 Flink Java 中創建 2 行的翻滾時間 window。 這必須基於 dateTime（TimeStamp3 數據類型）或 unixDateTime（BIGINT 數據類型）列。 我添加了兩個不同代碼版本的代碼。 我得到的錯誤放在代碼上方。

當我打印Table object 的數據類型時，我看到： |-- mID: INT |-- dateTime: TIMESTAMP(3) *ROWTIME* |-- mValue: DOUBLE |-- unixDateTime: BIGINT |-- mType: STRING

StreamExecutionEnvironment fsEnv = StreamExecutionEnvironment.getExecutionEnvironment();
        StreamTableEnvironment tableEnv = StreamTableEnvironment.create(fsEnv);   
        fsEnv.setStreamTimeCharacteristic(TimeCharacteristic.EventTime);

    TupleTypeInfo<Tuple5<Integer, Timestamp, Double, Long, String>> tupleType = new TupleTypeInfo<>(
        Types.INT(),
        Types.SQL_TIMESTAMP(),
        Types.DOUBLE(),
        Types.LONG(),
        Types.STRING());
        DataStream<Tuple5<Integer, Timestamp, Double, Long, String>> dsTuple =
                tableEnv.toAppendStream(HTable, tupleType);

//When I run below code I get this error: Caused by: java.lang.RuntimeException: Rowtime timestamp is null. Please make sure that a proper TimestampAssigner is defined and the stream environment uses the EventTime time characteristic.

Table table = tableEnv.fromDataStream(dsTuple, "mID, dateTime.rowtime, mValue, unixDateTime, mType");
   DataStream<Row> stream = tableEnv.toAppendStream(table, Row.class);
stream.print();

//When I run below code I get this error: Exception in thread "main" java.lang.UnsupportedOperationException: Event-time grouping windows on row intervals are currently not supported.

   Table table = tableEnv.fromDataStream(dsTuple, "mID, dateTime.rowtime, measurementValue, unixDateTime, measurementType")
.window(Tumble.over("2.rows")
.on("dateTime")
.as("a"))
.groupBy("a")
.select("AVG(mValue)");
    DataStream<Row> stream = tableEnv.toAppendStream(table, Row.class);
    stream.print();

Answer 1

對流表的基於時間的操作要求您明確告知 Flink 如何處理時間。 您需要閱讀文檔的相關部分。

您需要特別注意有關事件時間的部分。