[X] I searched in the issues and found nothing similar.
Motivation
In Flink jobs, the source and sink of paimon are both stateful. In our business scenarios, we may modify the parallelism, add sources, and replace source/sink tables, which can easily lead to problems where the state cannot be restored, such as Cannot map checkpoint/savepoint state for operator xxx to the new program.
For example, the following two operations cannot be restored from the Savepoint:
insert into table_A (select f1 from table_B) -> insert in table_A (select f1 from table_B union all select f1 from table_C)
insert into table_A select f1 from table_B -> insert into table_C select f1 from table_B
Therefore I want to set uid for source/sink operators to improve state compatibility when restoring from savepoint.
Search before asking
Motivation
In Flink jobs, the source and sink of paimon are both stateful. In our business scenarios, we may modify the parallelism, add sources, and replace source/sink tables, which can easily lead to problems where the state cannot be restored, such as
Cannot map checkpoint/savepoint state for operator xxx to the new program.For example, the following two operations cannot be restored from the Savepoint:
insert into table_A (select f1 from table_B)->insert in table_A (select f1 from table_B union all select f1 from table_C)insert into table_A select f1 from table_B->insert into table_C select f1 from table_BTherefore I want to set
uidfor source/sink operators to improve state compatibility when restoring from savepoint.Solution
Set uid for source/sink operators.
Anything else?
No response
Are you willing to submit a PR?