[
https://issues.apache.org/jira/browse/FLINK-36793?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
林健昌 updated FLINK-36793:
------------------------
Description: When Oracle takes incremental snapshots, there is a problem
with the logic of ChunkSplitter to fragment based on rowId, resulting in a very
large fragment, which will cause taskmanager memory overflow. (was:
oracle增量快照的时候，ChunkSplitter根据rowId去分片的逻辑有问题，导致有个分片会非常大，会导致taskmanager内存溢出的问题)
Summary: There is a problem with the block splitter logic of Oracle CDC
incremental snapshot, causing the split slice to be too large. (was: oracle
cdc的增量快照的块分割器逻辑有问题导致分割的片过大的问题)
> There is a problem with the block splitter logic of Oracle CDC incremental
> snapshot, causing the split slice to be too large.
> -----------------------------------------------------------------------------------------------------------------------------
>
> Key: FLINK-36793
> URL: https://issues.apache.org/jira/browse/FLINK-36793
> Project: Flink
> Issue Type: Improvement
> Components: Flink CDC
> Reporter: 林健昌
> Priority: Minor
> Original Estimate: 12h
> Remaining Estimate: 12h
>
> When Oracle takes incremental snapshots, there is a problem with the logic of
> ChunkSplitter to fragment based on rowId, resulting in a very large fragment,
> which will cause taskmanager memory overflow.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
