1996fanrui opened a new pull request, #29751: URL: https://github.com/apache/flink/pull/29751
## What is the purpose of the change Backport of #27767 and #28407 to release-2.2. `LocalRecoveryTest.testStateSizeIsConsideredForLocalRecoveryOnRestart` is flaky on this branch, which has neither commit. The test forces the executions to `RUNNING` while their deployment is still offloading the task restore, so deployment fails with "Cannot deploy ... during task restore offload", the job restarts and the checkpoint never registers. #28407 waits for the task deployment descriptors before forcing `RUNNING`; it builds on #27767, which adds `waitForAllTasksRunning`. Only test code changes. ## Verifying this change - `./mvnw -pl flink-runtime -am -Pskip-webui-build -Dtest='LocalRecoveryTest' -Dsurefire.failIfNoSpecifiedTests=false package` (JDK 17), 6 times: `Tests run: 1, Failures: 0, Errors: 0, Skipped: 0` each time - The same command with the test temporarily changed to `@RepeatedTest(500)` (JDK 17, change not committed): `Tests run: 500, Failures: 0, Errors: 0, Skipped: 0`, no "Cannot deploy" line in the output - `./mvnw -q -pl flink-runtime spotless:check checkstyle:check` (JDK 17): passes ## Difference from the original | Original commit | Conflict | Byte-identical | Changes | | --- | --- | --- | --- | | df4f8c48d60 | No | Yes | None | | e009616411c | No | Yes | None | ## Notes - The title is that of #27767, the original pull request; this pull request also carries #28407. - #29322 fixes a third race in this test (NPE in `PendingCheckpoint.finalizeCheckpoint`, also reported on this branch in the Jira). It is still open on master, so it is not included and that failure mode remains until it is backported. - No run was made on the base commit, so the local runs show that the test passes, not that the rate went down. The evidence for the race on this branch is the "Cannot deploy" lines in the CI logs linked from the Jira. cc @mukul-8 @RocMarshal @MartijnVisser ##### Was generative AI tooling used to co-author this PR? - [X] Yes (please specify the tool below) Generated-by: Claude Code (Claude Opus 5.5) 🤖 Generated with [Claude Code](https://claude.com/claude-code) -- 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]
