SEPURI-SAI-KRISHNA commented on PR #13288: URL: https://github.com/apache/seatunnel/pull/13288#issuecomment-6971179385
@DanielLeens Thanks, you are right: only `abortPrepare` drops the buffer. The sentence is narrowed in `cd16fa18a4` (EN and ZH). It now says that the sink drops the rows it still holds only when the engine aborts the writer, which Spark does after a failed write or checkpoint preparation and Zeta after a failed checkpoint preparation. In other cases a failed task closes the sink normally, and closing writes the buffered rows first, so rows received before the failure can reach the table. Either way the rows after the last checkpoint are written again on retry or restore, as the at-least-once note says. The `abortPrepare` Javadoc now also says "when preparing a checkpoint fails" for Zeta. I will merge dev once #13126 lands; the Windows JDK 11 lane runs again on that push. -- 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]
