flink-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From enrico canzonieri <ecanzoni...@gmail.com>
Subject Re: Tumbling window expects a time attribute for grouping in a stream environment
Date Thu, 25 May 2017 01:20:16 GMT
I solved this implementing a new Kafka09TableSource in my application. The
class I implemented extends both DefinedRowTimeAttribute and
DefinedProcTimeAttribute and it exposes the consumer so that I can assign
the timestamp extractor.

I'm not sure if this is the right approach, but if that's the case I wonder
if we could make those changes into KafkaTableSource to make it more
generic.

On Wed, May 24, 2017 at 12:23 PM, enrico canzonieri <ecanzonieri@gmail.com>
wrote:

> Hi Timo, thanks for your help!
>
> I tried to follow the examples in the tests but I still have the same
> issue.
> I changed my schema and added an additional field "rowtime". My schema now
> is:
> root
>  |-- rowtime: org.apache.flink.table.expressions.RowtimeAttribute(expr:
> GenericType<org.apache.flink.table.expressions.Expression>)
>  |-- time: Long
>  |-- host: String
>
> If I run the code:
> table.select('rowtime).toDataStream[Row].print()
> I get:
> RowtimeAttribute(1495580133000)
> RowtimeAttribute(1495580143000)
> RowtimeAttribute(1495580153000)
>
> But If I run:
> table.window(Tumble over 1.minutes on 'rowtime as 'w).groupBy('host,
> 'w).select('host)
> I still get the previous error:
> TumblingGroupWindow('w, 'rowtime, 60000.millis) is invalid: Tumbling
> window expects a time attribute for grouping in a stream environment.
>
> I'm using a Kafka09TableSource as data source, but it doesn't allow me to
> specify the timestamp assigner. I think the actual consumer is not exposed
> to the user so I cannot really call assignTimestampsAndWatermarks. May
> that be the problem? Should we expose that function so that we can assign
> timestamp and watermark to a TableSource?
>
> The time characteristic in the execution environment is set to EventTime
> in my code.
>
> Cheers,
> Enrico
>
> On Wed, May 24, 2017 at 2:08 AM, Timo Walther <twalthr@apache.org> wrote:
>
>> Hi Enrico,
>>
>> the docs of the 1.3-SNAPSHOT are a bit out of sync right now, but they
>> will be updated in the next days/1-2 weeks.
>>
>> We recently introduced so-called "time indicators". These are attributes
>> that correspond to Flink's time and watermarks. You declare a logical field
>> that represents Flink's internal time in a table program.
>>
>> In your example you need to append a "time.rowtime" or "time.proctime" to
>> your table schema definition.
>>
>> You can find some examples here:
>> https://github.com/apache/flink/blob/master/flink-libraries/
>> flink-table/src/test/scala/org/apache/flink/table/
>> runtime/datastream/TimeAttributesITCase.scala
>>
>> If you have further question, feel free to ask them. It helps us to
>> improve the documenation.
>>
>> Regards,
>> Timo
>>
>>
>>
>> Am 24.05.17 um 04:15 schrieb enrico canzonieri:
>>
>> Hi,
>> I'm trying to window and groupBy a stream using the table api, but I get
>> ValidationException in the windowing function.
>> Here is the relevant code:
>>
>> tableEnv.registerTableSource(schema.getName, src)
>> val table = tableEnv.scan(schema.getName)
>> val t = table.window(Tumble over 1.minutes on 'time as 'w).groupBy('host,
>> 'w).select('host)
>>
>> "time" is defined as Long in my schema. The error I get is:
>> Exception in thread "main" org.apache.flink.table.api.ValidationException:
>> TumblingGroupWindow('w, 'time, 60000.millis) is invalid: Tumbling window
>> expects a time attribute for grouping in a stream environment.
>>
>> I also tried to define a window that was using processing time, but what
>> described in the documentation "Tumble over 1.minutes as 'w"  doesn't
>> seem to work anymore. Specifically it seems that a window now always
>> expects the "on" call.
>>
>> Has anybody encountered this issue? I'm using Flink 1.3-SNAPSHOT.
>>
>> thanks
>>
>>
>>
>

Mime
View raw message