You are viewing a plain text version of this content. The canonical link for it is here.

Posted to builds@flink.apache.org by GitBox <gi...@apache.org> on 2026/09/16 11:57:37 UTC

[GH] (flink-cdc/iceberg-partition-update-fix): Workflow run "Flink CDC CI" is working again!

The GitHub Actions job "Flink CDC CI" on flink-cdc.git/iceberg-partition-update-fix has succeeded.
Run started by GitHub user dataengineervishal (triggered by ferenc-csaky).

Head commit for run:
c61adc3f626bbc32ced745a25bbe356167a860e2 / Vishal <vi...@gmail.com>
[FLINK-40640][iceberg] Fix duplicate rows when an UPDATE changes the partition value

Iceberg equality deletes are partition-scoped: a delete file only
applies to data files in the same partition with a lower sequence
number. RowDataUtils.convertDataChangeEventToRowData collapsed every
UPDATE into a single after-image row tagged RowKind.INSERT, discarding
the before-image entirely. When the update changed a partitioned
column's value, the resulting equality-delete was derived from and
routed to the new partition, so the stale row in the old partition was
never deleted.

Split UPDATE into two rows: a DELETE built from the before-image
(routed to and deleting from the old partition) followed by an INSERT
built from the after-image. IcebergWriter.write() now writes each row
returned by the (now list-returning) conversion.

Report URL: https://github.com/apache/flink-cdc/actions/runs/35067859351

With regards,
GitHub Actions via GitBox