DataStream 转换 Table 后事件时间不对

Fri, 13 May 2022 06:40:51 -0700 


DataStream 转换 Table 后事件时间怎么不对？是时区的问题吗？怎么解决？


```
DataStream<UserBehavior> sourceStream = env.fromElements(
        new UserBehavior(1001L, 3827899L, 2920476L, "pv", 1511713473000L, 
"2017-11-27 00:24:33"),
        new UserBehavior(1001L, 3745169L, 2891509L, "pv", 1511725471000L, 
"2017-11-27 03:44:31"),
        new UserBehavior(1001L, 1531036L, 2920476L, "pv", 1511733732000L, 
"2017-11-27 06:02:12"),
        new UserBehavior(1001L, 2266567L, 4145813L, "pv", 1511741471000L, 
"2017-11-27 08:11:11"),
        new UserBehavior(1001L, 2951368L, 1080785L, "pv", 1511750828000L, 
"2017-11-27 10:47:08"),
        new UserBehavior(1002L, 5002615L, 2520377L, "pv", 1511752985000L, 
"2017-11-27 11:23:05")
);


// 提取时间戳并分配 Watermark
DataStream<UserBehavior> behaviorStream = 
sourceStream.assignTimestampsAndWatermarks(
        
WatermarkStrategy.<UserBehavior>forBoundedOutOfOrderness(Duration.ofMinutes(1))
                .withTimestampAssigner(new 
SerializableTimestampAssigner<UserBehavior>() {
                    @Override
                    public long extractTimestamp(UserBehavior behavior, long 
recordTimestamp) {
                        return behavior.getTs();
                    }
                })
);


// 注册虚拟表
Table behaviorTable = tEnv.fromDataStream(behaviorStream, $("uid"), $("ts"), 
$("ts_row").rowtime());
tEnv.createTemporaryView("user_behavior", behaviorTable);


Table resultTable = tEnv.sqlQuery("SELECT uid, ts, ts_row FROM user_behavior");
DataStream<Row> resultStream = tEnv.toAppendStream(resultTable, Row.class);
resultStream.print();
```
输出结果：
```
3> +I[1001, 1511725471000, 2017-11-26T19:44:31]
4> +I[1001, 1511733732000, 2017-11-26T22:02:12]
3> +I[1002, 1511752985000, 2017-11-27T03:23:05]
2> +I[1001, 1511713473000, 2017-11-26T16:24:33]
2> +I[1001, 1511750828000, 2017-11-27T02:47:08]
1> +I[1001, 1511741471000, 2017-11-27T00:11:11]
```
1511725471000 -> 2017-11-27 03:44:31
目测是时区的问题

回复