Oracle NUMBER data type

[Roger Bieri]

Hi

I'm 'playing around' with Debezium & Oracle. Thanks to the Tutorial by Chris Cranford, it wasn't that hard to set everything up.

So I've created another user and table like 

CREATE TABLE source_app.cog_codegroup
(
cog_rowid NUMBER GENERATED BY DEFAULT ON NULL AS IDENTITY,
-- payload
cog_name VARCHAR(50),
cog_multi_value CHAR(1),
-- metadata of ETL-framework
etl_jobId INT DEFAULT 10000 NOT NULL,
    primary key (cog_rowid)
);

alter table source_app.cog_codegroup add supplemental log data (all) columns;

However the NUMBER types (and the INT) are received in the form of COG_ROWID":"Cw==" within the JSON strings. No matter the 'client' (consumer) be it kafka's consumer.sh or py-spark. It's obviously not a hex-code and it's probably case-sensitive.

I've also set decimal.handling.mode to precise, but that's default already according to the manual. So I wasn't feeling that lucky here.

In another test, I removed the IDENTITY and set a length like NUMBER(20, 0) which didn't help either.

What am I doing wrong?

[jiri.p...@gmail.com]

Hi,

please see https://debezium.io/documentation/faq/#how_to_retrieve_decimal_field_from_binary_representation and https://debezium.io/documentation/reference/2.4/connectors/oracle.html#oracle-property-decimal-handling-mode

J.