[
https://issues.apache.org/jira/browse/FLINK-31462?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17700658#comment-17700658
]
Ming Li commented on FLINK-31462:
---------------------------------
Hi, [~lzljs3620320] ,
For example, I consume two tables A and B. The data in table A is wrong after a
certain snapshot, so I need to re-import this part of the data. The data in
table B is correct, so there is no need to re-import.
When I want to backtrack, I hope that table A and table B can go back to the
same time period, instead of only backtracking table A, which will cause the
job to not see the data changes of table B in the corresponding time period,
only to see to the latest data.
One situation is that if the watermark is also derived from the data, there
will be no data in table B to participate in the calculation within the
corresponding time period. There may be problems in queries based on these
snapshots, because the data in table B does not actually participate in the
calculation.
> [Flink] Supports full calculation from specified snapshots in streaming mode
> ----------------------------------------------------------------------------
>
> Key: FLINK-31462
> URL: https://issues.apache.org/jira/browse/FLINK-31462
> Project: Flink
> Issue Type: Improvement
> Components: Table Store
> Reporter: Ming Li
> Priority: Major
>
> Currently, the table store provides a startup mode for incremental
> consumption from a specified snapshot in streaming mode. We can provide a
> startup mode for incremental consumption after full calculation from a
> specified snapshot.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
