litiliu opened a new pull request, #4609:
URL: https://github.com/apache/flink-cdc/pull/4609

   <!--
   Licensed to the Apache Software Foundation (ASF) under one
   or more contributor license agreements.  See the NOTICE file
   distributed with this work for additional information
   regarding copyright ownership.  The ASF licenses this file
   to you under the Apache License, Version 2.0 (the
   "License"); you may not use this file except in compliance
   with the License.  You may obtain a copy of the License at
   
     http://www.apache.org/licenses/LICENSE-2.0
   
   Unless required by applicable law or agreed to in writing,
   software distributed under the License is distributed on an
   "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
   KIND, either express or implied.  See the License for the
   specific language governing permissions and limitations
   under the License.
   -->
   
   ## What is the purpose of this pull request?
   
   Fix [FLINK-41312](https://issues.apache.org/jira/browse/FLINK-41312).
   
   When a partition is dropped or expires while a Fluss pipeline source is 
running, its splits and SDK subscriptions can remain active even though the 
logical table still exists. Repeated metadata requests for the absent partition 
can delay or stall consumption of surviving partitions.
   
   This change builds on the complete partition inventory introduced by 
[#4608](https://github.com/apache/flink-cdc/pull/4608) 
([FLINK-41313](https://issues.apache.org/jira/browse/FLINK-41313)) and retires 
already-active splits for absent partitions.
   
   This PR currently includes the shared changes from #4608 in its diff against 
`master`. After #4608 is merged, I will rebase this PR onto `master` so that 
only the running-reader retirement changes remain.
   
   ## Brief change log
   
   - Compare active reader splits with each successful complete partition 
inventory and request cleanup for absent partitions.
   - Execute precise partition/bucket unsubscribe and active or queued snapshot 
cancellation on the fetcher thread.
   - Return removed split IDs through the normal finished-split path while 
preserving surviving partitions and logical-table schema/resources.
   - Keep existing checkpoint serializer versions and split IDs. Same-name 
partition recreation with a new physical ID remains outside this change.
   
   ---
   
   ## Verifying this change
   
   Added ITs for partition-scoped SDK cleanup while preserving consumption of 
surviving partitions.
   
   ## Documentation
   
   - Does this pull request introduce a new feature? No; this is a bug fix.
   - If yes, how is the feature documented? Not applicable.
   
   ---
   
   ##### Was generative AI tooling used to co-author this PR?
   
   - [X] Yes — OpenAI Codex (GPT-6); GitHub Copilot (Claude Opus 5.5, read-only 
design discussion).
   
   Generated-by: OpenAI Codex (GPT-6); GitHub Copilot (Claude Opus 5.5)
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to