github-actions[bot] commented on code in PR #63079:
URL: https://github.com/apache/doris/pull/63079#discussion_r3240152247
##########
fe/fe-core/src/main/java/org/apache/doris/job/offset/jdbc/JdbcTvfSourceOffsetProvider.java:
##########
@@ -353,6 +344,7 @@ public void replayIfNeed(StreamingInsertJob job) throws
JobException {
Review Comment:
This no-committed-transaction restart branch restores `remainingSplits` from
`streaming_job_meta` but then returns before
`resumeCdcSplitProgressFromSplits()` runs. This is distinct from the existing
TVF replay cursor thread for the committed-offset path: here FE can restart
after `advanceSplits()` has persisted the first batch but before any TVF
snapshot task commits, so `currentOffset == null`. After the restored batch is
consumed, the single table is already present in
`remainingSplits`/`finishedSplits` and `cdcSplitProgress.currentSplittingTable`
is still null, so inherited `noMoreSplits()` reports true and the scheduler
stops fetching subsequent batches, truncating the snapshot. Please rebuild the
cursor before this branch returns (or fall through to the common
`resumeCdcSplitProgressFromSplits()` call) and add a restart test that restarts
after meta has splits but before the first committed txn.
--
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]
