azagrebin edited a comment on issue #7036: [FLINK-10531][e2e] Fix unstable TTL end-to-end test. URL: https://github.com/apache/flink/pull/7036#issuecomment-437038969 I agree that configuring custom processing time provider on the API side is more robust approach for tests. The change might be a bit more involving. I have created issues to consider it: FLINK-10830, FLINK-10831. One more option is to check if `currentTimeMillis` jumped back in `TtlVerifyUpdateFunction.performUpdate` relatively to the latest consumed value. If jump happened then just wait until `currentTimeMillis` returns the next value greater than the latest consumed value. This way the update never needs to be rejected for this reason and test restarted. There is still potential subtle problem (could be not really practical) if: - `timestampBeforeUpdate` gets increased value - Flink gets jumped back `currentTimeMillis` - lag happens and - `timestampAfterUpdate` gets again increased value. In this case, the time queried by Flink is completely out of test control and without plugging the time there is no way to sync it in Flink and test.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
