davidzollo opened a new pull request, #11944: URL: https://github.com/apache/seatunnel/pull/11944
## What does this PR do? This PR fixes a CDC enumerator gap where splits returned after reader recovery can stay pending even though the enumerator is already running. It also adds follow-up coverage around the engine restore stability path discussed in PR #11677. ## Why is this needed? `addSplitsBack(...)` re-queues restored splits, but before this change it did not immediately try to reassign them when the enumerator had already entered the running state. If a reader was already waiting for work, the job could stall until another scheduling trigger arrived. ## How was this fixed? - synchronize `addSplitsBack(...)` with the other scheduling entry points - immediately call `assignSplits()` when restored splits arrive after the enumerator is already running - add deterministic unit coverage for both pre-`run()` and post-`run()` add-splits-back paths - add a MySQL CDC E2E scenario that keeps parallel snapshot splitting active across repeated savepoint restores and validates end-to-end data consistency after multiple mutation rounds ## Validation - `./mvnw -pl seatunnel-connectors-v2/connector-cdc/connector-cdc-base,seatunnel-e2e/seatunnel-connector-v2-e2e/connector-cdc-mysql-e2e -am spotless:apply` - `./mvnw -pl seatunnel-e2e/seatunnel-connector-v2-e2e/connector-cdc-mysql-e2e spotless:apply` - compile and runtime validation will come from GitHub CI for this PR branch -- 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]
