On Mon, Jun 3, 2019 at 5:14 PM Yohei Onishi <[email protected]> wrote:

> Hi Nicolas,
>
> Are you running you job on Dataflow? According to GCP support, Dataflow
> currently does not support Schema Registry. But we can still use Schema
> Registry to serialize / deserialize your message using
> custom KafkaAvroSerializer.
>
> In my case I implemented my custom KafkaAvroDeserializer to deserialize
> Avro format message.
> You can implement custom KafkaAvroSerializer in a similar way.
>
> https://stackoverflow.com/questions/54755668/how-to-deserialising-kafka-avro-messages-using-apache-beam/55917157#55917157
>
> I hope this helps you.
>
> Yohei Onishi
>
>
> On Wed, May 29, 2019 at 5:44 PM Nicolas Delsaux <[email protected]>
> wrote:
>
>> Hello all
>>
>> I have a beam job that I use to read messages from RabbitMq t write them
>> in kafka.
>>
>> As of now, messages are read/written as JSON.
>>
>> Obviously, it's not that optimal storage, so i would like to transform
>> the messages to avro prior to write them in Kafka. I have the URL of a
>> schema registry I can use to store/get my schema.
>>
>
Can you simply have a ParDo that sits in between read and write steps and
perform the conversion (and also connect to schema registry) ?


>
>> But I see nowhere in Beam documentation how to transform my JSON into
>> Avro data (except by deserializing my JSON to a java class that i will
>> later transform into avro). Is that deserialization to class the only
>> way ? or is it possible to generate an avro generic record from my json
>> "directly" ?
>>
>> Once my avro data is generated, how can I write it to my Kafka topic ?
>>
>>
>> Thanks !
>>
>>

Reply via email to