linqigeng created FLINK-35547:
---------------------------------
Summary: Sources support HybridSource implements
Key: FLINK-35547
URL: https://issues.apache.org/jira/browse/FLINK-35547
Project: Flink
Issue Type: New Feature
Components: Flink CDC
Reporter: linqigeng
Since the specifications of slave instances are generally smaller than the
primary instance, it is more likely to restart when encountering performance
bottlenecks(e.g. CPU,memory,disk), and there is a certain delay with the
primary instance. In order not to affect the running Flink CDC jobs and improve
effectiveness, here is an idea:
Sources adapt HybridSouce, so we can do like:
* in the snapshot stage , readers fetch data from the slave instances
* after the snapshot phase ends, switches to reading binlog from the primary
instance
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
