Can you write the table if it doesn't exist?

[alberttwong]

Can you write the table if it doesn't exist? I am mapping dbserver1.inventory.customers topic to customer table

024-07-22 13:36:05 2024-07-22 20:36:05,820 INFO   ||  doLogic - Topic: [dbserver1.inventory.customers], Partition: [0], MinOffset: [0], MaxOffset: [3], Records: [4]   [com.clickhouse.kafka.connect.sink.processing.Processing]
2024-07-22 13:36:06 2024-07-22 20:36:06,120 ERROR  ||  Table [`default`.`customer`] does not exist - see docs for more details about table names and topic names.   [com.clickhouse.kafka.connect.sink.db.ClickHouseWriter]
2024-07-22 13:36:06 2024-07-22 20:36:06,122 WARN   ||  Deciding how to handle exception: Topic: [dbserver1.inventory.customers], Partition: [0], MinOffset: [0], MaxOffset: [3], (QueryId: [ce38f1fe-5171-46ca-927c-f510cb518c5d])   [com.clickhouse.kafka.connect.util.Utils]
2024-07-22 13:36:06 2024-07-22 20:36:06,122 ERROR  ||  Errors tolerance is disabled, wrapping exception: Topic: [dbserver1.inventory.customers], Partition: [0], MinOffset: [0], MaxOffset: [3], (QueryId: [ce38f1fe-5171-46ca-927c-f510cb518c5d])   [com.clickhouse.kafka.connect.util.Utils]

[Paultagoras]

We don't create the table if it doesn't exist, for at least a few reasons:

• We don't necessarily know how the data is structured
• There are optimizations that can happen that need more information from folks
• In theory the data is going to be accessed by other people/applications as well, which means there's likely some agreement around what will be there (that the connector wouldn't be aware of)

[alberttwong]

I get that but it was by a convenience. Given my other example of using debezium to move data from OLTP to OLAP, this would be a needed feature to just plug in a source and then the tables just show up in CH.