Skip to content

Caveats

Only iceberg V2 table supported

This connector writes data to Iceberg tables using the V2 specification. To optimize write performance, delete events are recorded in delete files, avoiding costly data file rewrites. While this approach significantly improves write performance, it can impact read performance, especially in upsert mode. However, in append mode, this performance trade-off is not applicable.

No automatic schema evolution

Full schema evolution, such as converting incompatible data types, is not currently supported. However, schema expansion, including adding new fields or expanding existing field data types, is supported. To enable this behavior, set the debezium.sink.iceberg.allow-field-addition configuration property to true.

Specific tables replication

By default, the Debezium connector will replicate all the tables in the database, resulting in unnecessary load. To avoid replicating tables you don't need, configure the debezium.source.table.include.list property to specify the exact tables to replicate. This will streamline your data pipeline and reduce the overhead. For more details on this configuration, refer to the Debezium server source documentation.

AWS S3 credentials

You can setup aws credentials in the following ways:

  • Option 1: use debezium.sink.iceberg.fs.s3a.access.key and debezium.sink.iceberg.fs.s3a.secret.key in application.properties
  • Option 2: inject credentials to environment variables AWS_ACCESS_KEY and AWS_SECRET_ACCESS_KEY
  • Option 3: setup proper HADOOP_HOME env then add s3a configuration into core-site.xml, more information can be found here.