dybyte opened a new issue, #13017: URL: https://github.com/apache/seatunnel/issues/13017
### Search before asking - [x] I had searched in the [feature](https://github.com/apache/seatunnel/issues?q=is%3Aissue+label%3A%22Feature%22) and found no similar feature requirement. ### Description `engine_checkpoint-id-map` holds each pipeline's checkpoint ID counter. It matches `engine*`, so every checkpoint appends a write to the IMap storage, and all of it is replayed on restart. #12022 measured 492 MB (~60 MB/day) with 40 streaming jobs over 8 days. Since #12387, restore no longer needs the persisted value: `CheckpointManager` reads the latest completed checkpoint from checkpoint storage and calls `setCount(checkpointId + 1)`. Without the persisted value, the counter starts at 1 after a full restart and is then overwritten from the latest completed checkpoint. It only stays at 1 when a pipeline has no completed checkpoint, where nothing was committed under those IDs and 2PC sinks (e.g. Doris, Kafka) derive their transaction IDs from their own state. ### Usage Scenario Never persist `engine_checkpoint-id-map`, like `engine_runningJobMetrics` and `engine_checkpoint_monitor` (#13013). After a full cluster restart, the counter is rebuilt from checkpoint storage. ### Related issues #12022, #12387, #13013 ### Are you willing to submit a PR? - [x] Yes I am willing to submit a PR! ### Code of Conduct - [x] I agree to follow this project's [Code of Conduct](https://www.apache.org/foundation/policies/conduct) -- 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]
