When the schema of your upstream source changes (i.e. adding or removing columns), you need to modify the schema of the corresponding source or table in RisingWave to ensure they are aligned. If the schema registry is used to create a source or table, you cannot modify the schema of the source or table.

​
Add a column

To add a column to a source, use this command: 
ALTER SOURCE <source_name> ADD COLUMN <column_name> <column_type>;
Similarly, to add a column to a table, use this command: 
ALTER TABLE <table_name> ADD COLUMN <column_name> <column_type>;
For details about these two commands, see ALTER SOURCE and ALTER TABLE.
Note that you cannot add a primary key column to a source or table in RisingWave. To modify the primary key of a source or table, you need to recreate the table.
When you add a column to a source or table, the new column is not automatically picked up in a downstream materialized view. For example, if you have a materialized view mv1 that is created with this statement: 
CREATE MATERIALIZED VIEW mv1 AS SELECT * FROM my_kafka_source;
When you add a column to my_kafka_source: 
ALTER SOURCE my_kafka_source ADD COLUMN new_col INTEGER;
The materialized view mv1 does not contain the new column new_col. If you want to pick up this new column, you’ll need to create a new materialized view.

​
Remove a column

You cannot remove a column from a source in RisingWave. If you intend to remove a column from a source, you’ll need to drop the source and create the source again. If a column of a table is referenced in a materialized view, you cannot remove the column. This is to ensure data consistency and prevent disruptions to downstream data processing. To remove a column from a table, use this command: 
ALTER TABLE table_name DROP COLUMN column_name;

​
Refresh the schema registry

​
Source

At present, combined with the ALTER SOURCE command, you can refresh the schema registry of a source by refilling its FORMAT and ENCODE options. The syntax is: 
ALTER SOURCE source_name FORMAT data_format ENCODE data_encode [ (
    message='message',
    schema.location='location', ...) ];
Here is a simple example. Let’s assume we have a source as follows: 
-- Create a source.
CREATE SOURCE src_user WITH (
    connector = 'kafka',
    topic = 'sr_pb_test',
    properties.bootstrap.server = 'message_queue:29092',
    scan.startup.mode = 'earliest'
)
FORMAT PLAIN ENCODE PROTOBUF(
    schema.registry = 'http://message_queue:8081',
    message = 'test.User');
Then you can refresh the schema registry by the following command: 
ALTER SOURCE src_user FORMAT PLAIN ENCODE PROTOBUF(
    schema.registry = 'http://message_queue:8081',
    message = 'test.UserWithMoreFields'
);
Currently, it is not supported to modify the data_format and data_encode. Furthermore, when refreshing the schema registry of a source, it is not allowed to drop columns or change types.
In addition, when the FORMAT and ENCODE options are not changed, the REFRESH SCHEMA clause of ALTER SOURCE can also be used to refresh the schema of a source. 
ALTER SOURCE source_name REFRESH SCHEMA;
For example, assume we have a source as follows: Create a source 
CREATE SOURCE src_user WITH (
    connector = 'kafka',
    topic = 'sr_pb_test',
    properties.bootstrap.server = 'message_queue:29092',
    scan.startup.mode = 'earliest'
)
FORMAT PLAIN ENCODE PROTOBUF(
    schema.registry = 'http://message_queue:8081',
    message = 'test.User'
);
Then we can refresh its schema with the following statement: Refresh schema 
ALTER SOURCE src_user REFRESH SCHEMA;

​
Table

Similarly, you can use the following statement to refresh the schema of a table with connectors. For more details, see ALTER TABLE. Refresh schema of table 
ALTER TABLE src_user REFRESH SCHEMA;
If a downstream fragment references a column that is either missing or has undergone a type change in the updated schema, the command will be declined.

​
See also