Broadcasting to multiple operators

[relink]

Hello.
I am having an issue with a Flink 1.8 pipeline when trying to consume broadcast state across multiple operators.  I currently
have a working pipeline that looks like the following:

records
    .assignTimestampsAndWatermarks(
        new BoundedOutOfOrdernessGenerator(
            Long.parseLong(properties.getProperty(("flinkMaxOutOfOrderness")))))
    .keyBy(new ApplicationNameKeySelector())
    .window(
        TumblingEventTimeWindows.of(
            Time.seconds(Long.parseLong(properties.getProperty("flinkWindow")))))
    .aggregate(new Aggregator())
    .connect(configurationBroadcastStream)
    .process(excluder)
    .addSink(KinesisProducer.createSinkFromStaticConfig(properties));

* records are a FlinkKafkaConsumer stream
* configurationBroadcastStream is a FlinkKafkaConsumer

* aggregator is an AggregateFunction
* filter is a BroadcastProcessFunction

I now have requirements to filter out transactions at the beginning of the pipeline using the same broadcast stream I am consuming towards the end of the pipeline. I updated the pipeline to look like this:

records
    .assignTimestampsAndWatermarks(
        new BoundedOutOfOrdernessGenerator(
            Long.parseLong(properties.getProperty(("flinkMaxOutOfOrderness")))))
    .connect(configurationBroadcastStream) **new**
    .process(filter) **new**
    .keyBy(new ApplicationNameKeySelector())
    .window(
        TumblingEventTimeWindows.of(
            Time.seconds(Long.parseLong(properties.getProperty("flinkWindow")))))
    .aggregate(new Aggregator())
    .connect(configurationBroadcastStream)
    .process(excluder)
    .addSink(KinesisProducer.createSinkFromStaticConfig(properties));

* records are a FlinkKafkaConsumer stream
* configurationBroadcastStream is a FlinkKafkaConsumer

* aggregator is an AggregateFunction
* excluder is a BroadcastProcessFunction

With this change, the aggregated records are not making it into the excluder process.

1. The aggregator add is working. I can see this in the logs.

2. The aggregator getResult is never called. This makes me think this is a window issue.

3. Both processBroadcastElement methods from the two broadcast functions are working and 

     retrieving the broadcasted state. I see this in logging.

4. The pipeline definitely worked prior to me adding in the second .connect and .process at the beginning of the pipeline.

5. I have considered creating a new record object from the new process(filter) that contains the config retrieved from the broadcast stream along with the transactions and passing that down the pipeline but that is really not desirable.

Any ideas on what might be going on here?

Thanks!

Roger

[David Anderson-4]

This is a watermarking issue. Whenever an operator has two or more input streams, its watermark is the minimum of watermarks of the incoming streams. In this case your broadcast stream doesn't have a watermark generator, so it is preventing the watermarks from advancing. This in turn is preventing the windows from being triggered.

You should call assignTimestampsAndWatermarks on the broadcast stream. If time is irrelevant for this stream, you could do something like this:

public static class ConfigStreamAssigner implements AssignerWithPeriodicWatermarks<T> {
      @Nullable
      @Override
      public Watermark getCurrentWatermark() {
         return Watermark.MAX_WATERMARK;
      }

      @Override
      public long extractTimestamp(T element, long previousElementTimestamp) {
         return 0;
      }
}

By setting the watermark for this stream to MAX_WATERMARK, you are effectively removing this stream's watermarks from consideration.

Regards,
David

On Fri, Mar 5, 2021 at 5:48 PM Roger <[hidden email]> wrote:

Hello.
I am having an issue with a Flink 1.8 pipeline when trying to consume broadcast state across multiple operators.  I currently
have a working pipeline that looks like the following:

records
    .assignTimestampsAndWatermarks(
        new BoundedOutOfOrdernessGenerator(
            Long.parseLong(properties.getProperty(("flinkMaxOutOfOrderness")))))
    .keyBy(new ApplicationNameKeySelector())
    .window(
        TumblingEventTimeWindows.of(
            Time.seconds(Long.parseLong(properties.getProperty("flinkWindow")))))
    .aggregate(new Aggregator())
    .connect(configurationBroadcastStream)
    .process(excluder)
    .addSink(KinesisProducer.createSinkFromStaticConfig(properties));

* records are a FlinkKafkaConsumer stream
* configurationBroadcastStream is a FlinkKafkaConsumer

* aggregator is an AggregateFunction
* filter is a BroadcastProcessFunction

I now have requirements to filter out transactions at the beginning of the pipeline using the same broadcast stream I am consuming towards the end of the pipeline. I updated the pipeline to look like this:

records
    .assignTimestampsAndWatermarks(
        new BoundedOutOfOrdernessGenerator(
            Long.parseLong(properties.getProperty(("flinkMaxOutOfOrderness")))))
    .connect(configurationBroadcastStream) **new**
    .process(filter) **new**
    .keyBy(new ApplicationNameKeySelector())
    .window(
        TumblingEventTimeWindows.of(
            Time.seconds(Long.parseLong(properties.getProperty("flinkWindow")))))
    .aggregate(new Aggregator())
    .connect(configurationBroadcastStream)
    .process(excluder)
    .addSink(KinesisProducer.createSinkFromStaticConfig(properties));

* records are a FlinkKafkaConsumer stream
* configurationBroadcastStream is a FlinkKafkaConsumer

* aggregator is an AggregateFunction
* excluder is a BroadcastProcessFunction

With this change, the aggregated records are not making it into the excluder process.

1. The aggregator add is working. I can see this in the logs.

2. The aggregator getResult is never called. This makes me think this is a window issue.

3. Both processBroadcastElement methods from the two broadcast functions are working and 

     retrieving the broadcasted state. I see this in logging.

4. The pipeline definitely worked prior to me adding in the second .connect and .process at the beginning of the pipeline.

5. I have considered creating a new record object from the new process(filter) that contains the config retrieved from the broadcast stream along with the transactions and passing that down the pipeline but that is really not desirable.

Any ideas on what might be going on here?

Thanks!

Roger

[relink]

Hey David.

Thank you very much for your response. This is making sense now. It was confusing because I was able to use the Broadcast stream prior to adding the second stream. However, now I realize that this part of the pipeline occurs after the windowing so I'm not affected the same way. This is definitely going to help fix my problem. 

On Fri, Mar 5, 2021 at 12:33 PM David Anderson <[hidden email]> wrote:

This is a watermarking issue. Whenever an operator has two or more input streams, its watermark is the minimum of watermarks of the incoming streams. In this case your broadcast stream doesn't have a watermark generator, so it is preventing the watermarks from advancing. This in turn is preventing the windows from being triggered.

You should call assignTimestampsAndWatermarks on the broadcast stream. If time is irrelevant for this stream, you could do something like this:

public static class ConfigStreamAssigner implements AssignerWithPeriodicWatermarks<T> {
      @Nullable
      @Override
      public Watermark getCurrentWatermark() {
         return Watermark.MAX_WATERMARK;
      }

      @Override
      public long extractTimestamp(T element, long previousElementTimestamp) {
         return 0;
      }
}

By setting the watermark for this stream to MAX_WATERMARK, you are effectively removing this stream's watermarks from consideration.

Regards,
David

On Fri, Mar 5, 2021 at 5:48 PM Roger <[hidden email]> wrote:

Hello.
I am having an issue with a Flink 1.8 pipeline when trying to consume broadcast state across multiple operators.  I currently
have a working pipeline that looks like the following:

records
    .assignTimestampsAndWatermarks(
        new BoundedOutOfOrdernessGenerator(
            Long.parseLong(properties.getProperty(("flinkMaxOutOfOrderness")))))
    .keyBy(new ApplicationNameKeySelector())
    .window(
        TumblingEventTimeWindows.of(
            Time.seconds(Long.parseLong(properties.getProperty("flinkWindow")))))
    .aggregate(new Aggregator())
    .connect(configurationBroadcastStream)
    .process(excluder)
    .addSink(KinesisProducer.createSinkFromStaticConfig(properties));

* records are a FlinkKafkaConsumer stream
* configurationBroadcastStream is a FlinkKafkaConsumer

* aggregator is an AggregateFunction
* filter is a BroadcastProcessFunction

I now have requirements to filter out transactions at the beginning of the pipeline using the same broadcast stream I am consuming towards the end of the pipeline. I updated the pipeline to look like this:

records
    .assignTimestampsAndWatermarks(
        new BoundedOutOfOrdernessGenerator(
            Long.parseLong(properties.getProperty(("flinkMaxOutOfOrderness")))))
    .connect(configurationBroadcastStream) **new**
    .process(filter) **new**
    .keyBy(new ApplicationNameKeySelector())
    .window(
        TumblingEventTimeWindows.of(
            Time.seconds(Long.parseLong(properties.getProperty("flinkWindow")))))
    .aggregate(new Aggregator())
    .connect(configurationBroadcastStream)
    .process(excluder)
    .addSink(KinesisProducer.createSinkFromStaticConfig(properties));

* records are a FlinkKafkaConsumer stream
* configurationBroadcastStream is a FlinkKafkaConsumer

* aggregator is an AggregateFunction
* excluder is a BroadcastProcessFunction

With this change, the aggregated records are not making it into the excluder process.

1. The aggregator add is working. I can see this in the logs.

2. The aggregator getResult is never called. This makes me think this is a window issue.

3. Both processBroadcastElement methods from the two broadcast functions are working and 

     retrieving the broadcasted state. I see this in logging.

4. The pipeline definitely worked prior to me adding in the second .connect and .process at the beginning of the pipeline.

5. I have considered creating a new record object from the new process(filter) that contains the config retrieved from the broadcast stream along with the transactions and passing that down the pipeline but that is really not desirable.

Any ideas on what might be going on here?

Thanks!

Roger

[relink]

Confirmed. This worked! 

Thanks!

Roger

On Fri, Mar 5, 2021 at 12:41 PM Roger <[hidden email]> wrote:

Hey David.

Thank you very much for your response. This is making sense now. It was confusing because I was able to use the Broadcast stream prior to adding the second stream. However, now I realize that this part of the pipeline occurs after the windowing so I'm not affected the same way. This is definitely going to help fix my problem. 

On Fri, Mar 5, 2021 at 12:33 PM David Anderson <[hidden email]> wrote:

This is a watermarking issue. Whenever an operator has two or more input streams, its watermark is the minimum of watermarks of the incoming streams. In this case your broadcast stream doesn't have a watermark generator, so it is preventing the watermarks from advancing. This in turn is preventing the windows from being triggered.

You should call assignTimestampsAndWatermarks on the broadcast stream. If time is irrelevant for this stream, you could do something like this:

public static class ConfigStreamAssigner implements AssignerWithPeriodicWatermarks<T> {
      @Nullable
      @Override
      public Watermark getCurrentWatermark() {
         return Watermark.MAX_WATERMARK;
      }

      @Override
      public long extractTimestamp(T element, long previousElementTimestamp) {
         return 0;
      }
}

By setting the watermark for this stream to MAX_WATERMARK, you are effectively removing this stream's watermarks from consideration.

Regards,
David

On Fri, Mar 5, 2021 at 5:48 PM Roger <[hidden email]> wrote:

Hello.
I am having an issue with a Flink 1.8 pipeline when trying to consume broadcast state across multiple operators.  I currently
have a working pipeline that looks like the following:

records
    .assignTimestampsAndWatermarks(
        new BoundedOutOfOrdernessGenerator(
            Long.parseLong(properties.getProperty(("flinkMaxOutOfOrderness")))))
    .keyBy(new ApplicationNameKeySelector())
    .window(
        TumblingEventTimeWindows.of(
            Time.seconds(Long.parseLong(properties.getProperty("flinkWindow")))))
    .aggregate(new Aggregator())
    .connect(configurationBroadcastStream)
    .process(excluder)
    .addSink(KinesisProducer.createSinkFromStaticConfig(properties));

* records are a FlinkKafkaConsumer stream
* configurationBroadcastStream is a FlinkKafkaConsumer

* aggregator is an AggregateFunction
* filter is a BroadcastProcessFunction

I now have requirements to filter out transactions at the beginning of the pipeline using the same broadcast stream I am consuming towards the end of the pipeline. I updated the pipeline to look like this:

records
    .assignTimestampsAndWatermarks(
        new BoundedOutOfOrdernessGenerator(
            Long.parseLong(properties.getProperty(("flinkMaxOutOfOrderness")))))
    .connect(configurationBroadcastStream) **new**
    .process(filter) **new**
    .keyBy(new ApplicationNameKeySelector())
    .window(
        TumblingEventTimeWindows.of(
            Time.seconds(Long.parseLong(properties.getProperty("flinkWindow")))))
    .aggregate(new Aggregator())
    .connect(configurationBroadcastStream)
    .process(excluder)
    .addSink(KinesisProducer.createSinkFromStaticConfig(properties));

* records are a FlinkKafkaConsumer stream
* configurationBroadcastStream is a FlinkKafkaConsumer

* aggregator is an AggregateFunction
* excluder is a BroadcastProcessFunction

With this change, the aggregated records are not making it into the excluder process.

1. The aggregator add is working. I can see this in the logs.

2. The aggregator getResult is never called. This makes me think this is a window issue.

3. Both processBroadcastElement methods from the two broadcast functions are working and 

     retrieving the broadcasted state. I see this in logging.

4. The pipeline definitely worked prior to me adding in the second .connect and .process at the beginning of the pipeline.

5. I have considered creating a new record object from the new process(filter) that contains the config retrieved from the broadcast stream along with the transactions and passing that down the pipeline but that is really not desirable.

Any ideas on what might be going on here?

Thanks!

Roger

[David Anderson-4]

Glad to hear it! Thanks for letting us know.

David

On Fri, Mar 5, 2021 at 10:22 PM Roger <[hidden email]> wrote:

Confirmed. This worked! 

Thanks!

Roger

On Fri, Mar 5, 2021 at 12:41 PM Roger <[hidden email]> wrote:

Hey David.

Thank you very much for your response. This is making sense now. It was confusing because I was able to use the Broadcast stream prior to adding the second stream. However, now I realize that this part of the pipeline occurs after the windowing so I'm not affected the same way. This is definitely going to help fix my problem. 

On Fri, Mar 5, 2021 at 12:33 PM David Anderson <[hidden email]> wrote:

This is a watermarking issue. Whenever an operator has two or more input streams, its watermark is the minimum of watermarks of the incoming streams. In this case your broadcast stream doesn't have a watermark generator, so it is preventing the watermarks from advancing. This in turn is preventing the windows from being triggered.

You should call assignTimestampsAndWatermarks on the broadcast stream. If time is irrelevant for this stream, you could do something like this:

public static class ConfigStreamAssigner implements AssignerWithPeriodicWatermarks<T> {
      @Nullable
      @Override
      public Watermark getCurrentWatermark() {
         return Watermark.MAX_WATERMARK;
      }

      @Override
      public long extractTimestamp(T element, long previousElementTimestamp) {
         return 0;
      }
}

By setting the watermark for this stream to MAX_WATERMARK, you are effectively removing this stream's watermarks from consideration.

Regards,
David

On Fri, Mar 5, 2021 at 5:48 PM Roger <[hidden email]> wrote:

Hello.
I am having an issue with a Flink 1.8 pipeline when trying to consume broadcast state across multiple operators.  I currently
have a working pipeline that looks like the following:

records
    .assignTimestampsAndWatermarks(
        new BoundedOutOfOrdernessGenerator(
            Long.parseLong(properties.getProperty(("flinkMaxOutOfOrderness")))))
    .keyBy(new ApplicationNameKeySelector())
    .window(
        TumblingEventTimeWindows.of(
            Time.seconds(Long.parseLong(properties.getProperty("flinkWindow")))))
    .aggregate(new Aggregator())
    .connect(configurationBroadcastStream)
    .process(excluder)
    .addSink(KinesisProducer.createSinkFromStaticConfig(properties));

* records are a FlinkKafkaConsumer stream
* configurationBroadcastStream is a FlinkKafkaConsumer

* aggregator is an AggregateFunction
* filter is a BroadcastProcessFunction

I now have requirements to filter out transactions at the beginning of the pipeline using the same broadcast stream I am consuming towards the end of the pipeline. I updated the pipeline to look like this:

records
    .assignTimestampsAndWatermarks(
        new BoundedOutOfOrdernessGenerator(
            Long.parseLong(properties.getProperty(("flinkMaxOutOfOrderness")))))
    .connect(configurationBroadcastStream) **new**
    .process(filter) **new**
    .keyBy(new ApplicationNameKeySelector())
    .window(
        TumblingEventTimeWindows.of(
            Time.seconds(Long.parseLong(properties.getProperty("flinkWindow")))))
    .aggregate(new Aggregator())
    .connect(configurationBroadcastStream)
    .process(excluder)
    .addSink(KinesisProducer.createSinkFromStaticConfig(properties));

* records are a FlinkKafkaConsumer stream
* configurationBroadcastStream is a FlinkKafkaConsumer

* aggregator is an AggregateFunction
* excluder is a BroadcastProcessFunction

With this change, the aggregated records are not making it into the excluder process.

1. The aggregator add is working. I can see this in the logs.

2. The aggregator getResult is never called. This makes me think this is a window issue.

3. Both processBroadcastElement methods from the two broadcast functions are working and 

     retrieving the broadcasted state. I see this in logging.

4. The pipeline definitely worked prior to me adding in the second .connect and .process at the beginning of the pipeline.

5. I have considered creating a new record object from the new process(filter) that contains the config retrieved from the broadcast stream along with the transactions and passing that down the pipeline but that is really not desirable.

Any ideas on what might be going on here?

Thanks!

Roger