MOBIN created FLINK-36632:
-----------------------------
Summary: When scan.startup.mode is not initial and restart with
savepoint, newly add-table will not synchronize snapshot data
Key: FLINK-36632
URL: https://issues.apache.org/jira/browse/FLINK-36632
Project: Flink
Issue Type: Bug
Components: Flink CDC
Affects Versions: cdc-3.2.0
Environment: cdc branch： master
Flink 1.19
Reporter: MOBIN
Minimum repetition step：
1. pipeline.yml such as
{code:java}
source:
type: mysql
...
tables: test_db.test1
schema-change.enabled: true
scan.startup.mode: timestamp
scan.startup.timestamp-millis: 1730096827000
scan.newly-added-table.enabled: true
sink:
type: paimon
....
pipeline:
name: MySQL to Paimon Test Pipeline
parallelism: 1
yarn.application.queue: root.flink-minor
yarn.tags: cdc-paimon-pipeline {code}
2. _cancel_ job with _savepoint_
_3. add new table:_ {_}tables: test_db.test1 -> tables:
test_db.test1,{_}{_}test_db.test2{_}
4. restart job with savepoint
The historical snapshot data of test_db.test2 is not synchronized. Only the
data after [scan.startup.timestamp-millis: 1730096827000] is synchronized.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
