mymeiyi opened a new pull request, #67661: URL: https://github.com/apache/doris/pull/67661
Related PR: #65196 Problem Summary: A rollup job in WAITING_TXN checks and aborts failed conflict transactions before proceeding. If a transaction is already committed, becomes visible, or is cleaned up before the abort completes, abortTransaction() can throw UserException. RollupJobV2 propagates this as AlterCancelException, causing the entire rollup job to be cancelled. Restarting FE with committed but not yet visible transactions reproduces this failure (the txn_insert_restart_fe_with_schema_change regression case can produce this bug). Apply the retry behavior introduced for schema change in #65196 to rollup jobs: log the abort failure, keep the job in WAITING_TXN, and re-query transaction state in the next scheduler round. Also exclude committed and final transactions from the shared failed-transaction selector to avoid unnecessary abort attempts. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
