Mao Jiayi created FLINK-40130:
---------------------------------
Summary: Fix missing primary key type compatibility check of
Paimon sink
Key: FLINK-40130
URL: https://issues.apache.org/jira/browse/FLINK-40130
Project: Flink
Issue Type: Bug
Components: Flink CDC
Reporter: Mao Jiayi
When writing to an existing Paimon table, the Paimon sink may continue with
record coercion even if the upstream primary key column type is incompatible
with the primary key column type of the existing Paimon table.
The issue happens in {*}{{BucketAssignOperator}}{*}. The current implementation
compares the upstream schema with the Paimon physical table schema and attempts
to coerce records when the schemas are different. However, it does not
explicitly validate whether primary key columns are type-compatible before
coercing records.
This issue may not be exposed when the primary key types are identical or
safely compatible, but it can fail or behave unexpectedly when the existing
Paimon table has a primary key type such as *{{INT}}* while the upstream schema
provides the same primary key column as {*}{{STRING}}{*}. In such cases, the
sink should reject the incompatible primary key type early with a clear error
instead of continuing into the record conversion path.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
