Thanks for that.  Still digging; stay tuned...

On Fri, Dec 16, 2022 at 4:32 PM Austin Tao <[email protected]> wrote:

> Hi Paul,
>
> Thank you for your quick reply. Keeping some privacy, attached is the
> configuration used for the processor, and the "AvroRecordSetWriter - key"
> I'm using.
>
> On Friday, December 16, 2022 at 03:58:10 p.m. EST, Paul Grey <
> [email protected]> wrote:
>
>
> Hi.  I worked that feature; I'll have a look.
>
> When the "Publish Strategy" is set to "Use Wrapper", my recollection is
> that a second record schema is needed for the key, depending on the
> processor configuration as a whole.  Are you able to share that processor
> configuration?
>
> Thanks
>
> On Fri, Dec 16, 2022 at 3:46 PM Austin Tao via users <
> [email protected]> wrote:
>
> To whom it may concern,
>
> Hello, I would like to report an issue for Nifi. But, following the new
> Jira Guidelines, I would therefore like to request that an account for ASF
> Jira in order to create a ticket.
>
> In regards to the bug, using Nifi 1.19.1 I would like to send a tombstone
> message (null payload) to Kafka and using the Confluent JDBC sink
> connector
> <https://docs.confluent.io/kafka-connectors/jdbc/current/sink-connector/sink_config_options.html>
>  (with
> delete.enabled=true) to delete a record in our Postgres database. I believe
> as of Nifi 1.19, PublishKafkaRecord_2_6 now supports the '*Publish
> Strategy*: Use Wrapper' feature which allows setting the Kafka message
> key and value (Primary Key as the Kafka key, null for the Kafka value). For
> the Record Key Writer, I'm using an AvroRecordSetWriter to validate and
> serialize the key against the confluent schema registry (*Schema Write
> Strategy*: Confluent Schema Registry Reference, *Schema Access Strategy*:
> Use 'Schema Name' Property) but sending the message I come across the error:
> *PublishKafkaRecord_2_6[id=XXX] Failed to send FlowFile[filename=XXX] to
> Kafka: org.apache.nifi.processor.exception.ProcessException: Could not
> determine the Avro Schema to use for writing the content*
>
> *- Caused by: org.apache.nifi.schema.access.SchemaNotFoundException:
> Cannot write Confluent Schema Registry Reference because the Schema
> Identifier is not known*
>
>
> I can confirm the configuration for the for the AvroRecordSetWriter,
> ConfluentSchemaRegistry controllers, and PublishKafkaRecord processor are
> all configured correctly, as I can send the Kafka message just fine using
> the default *Publish Strategy* (Use Content as Record Value). It only
> occurs using Use Wrapper, and the ConfluentSchemaRegistry.
>
> A workaround that has worked was for using JsonRecordSetWriter w/ embedded
> JSON schemas, but it would be nice to continue using our Avro Schema
> Registry for this.
>
> I'd appreciate if someone had any advice or experience with this issue,
> otherwise I'd like to log an issue in JIRA.
>
> Thank you,
> Austin Tao
>
>

Reply via email to