[
https://issues.apache.org/jira/browse/FLINK-38184?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hongshun Wang updated FLINK-38184:
----------------------------------
Description:
When I have a big Postgres source table(1 billion data), then the checkpoint
will cost multiple minutes which will block the whole reading. The main cost is
org.apache.flink.core.memory.DataOutputSerializer#getCopyOfBuffer.
!image-2025-08-04-11-03-18-074.png!
To be host, it's a no meaning operation. For one split, it only need once
out.getCopyOfBuffer rather than multiple times.
was:
When I have a big Postgres source table(1 billion data), then the checkpoint
will cost multiple minutes which will block the whole reading. The main cost is
org.apache.flink.core.memory.DataOutputSerializer#getCopyOfBuffer.
!image-2025-08-04-11-03-18-074.png!
To behost, it's a no meaning operation. For one split, it only need once
out.getCopyOfBuffer rather than multiple times.
> CDC no need to getCopyOfBuffer for each split info.
> ---------------------------------------------------
>
> Key: FLINK-38184
> URL: https://issues.apache.org/jira/browse/FLINK-38184
> Project: Flink
> Issue Type: Improvement
> Components: Flink CDC
> Affects Versions: cdc-3.4.0
> Reporter: Hongshun Wang
> Priority: Major
> Labels: pull-request-available
> Fix For: cdc-3.5.0
>
> Attachments: image-2025-08-04-11-03-18-074.png
>
>
> When I have a big Postgres source table(1 billion data), then the checkpoint
> will cost multiple minutes which will block the whole reading. The main cost
> is org.apache.flink.core.memory.DataOutputSerializer#getCopyOfBuffer.
> !image-2025-08-04-11-03-18-074.png!
> To be host, it's a no meaning operation. For one split, it only need once
> out.getCopyOfBuffer rather than multiple times.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
