[
https://issues.apache.org/jira/browse/FLINK-35036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17834749#comment-17834749
]
Biao Geng commented on FLINK-35036:
-----------------------------------
Hi [~fly365], according to the attached screenshot, the failure is caused by a
timeout in flink client side.
IIUC, in the full volume phase of a flink cdc job, it needs to process lots of
data and typically due to the back pressure, the state may be much larger than
the incremental phase(you can check the state size in flink's web ui).
As a result, it would take longer time for the flink to complete the savepoint.
The client's [default timeout is
60s|https://nightlies.apache.org/flink/flink-docs-master/docs/deployment/config/#client-timeout],
so maybe you can increase the value to see if the savepoint can succeed.
> Flink CDC Job cancel with savepoint failed
> ------------------------------------------
>
> Key: FLINK-35036
> URL: https://issues.apache.org/jira/browse/FLINK-35036
> Project: Flink
> Issue Type: Bug
> Components: Flink CDC
> Environment: Flink 1.15.2
> Flink CDC 2.4.2
> Oracle 19C
> Doris 2.0.3
> Reporter: Fly365
> Priority: Major
> Attachments: image-2024-04-07-17-35-23-136.png
>
>
> With the Flink CDC job, I want oracle data to doris, in the snapshot，canel
> the Flink CDC Job with savepoint,the job cancel failed.
> 使用Flink CDC,将Oracle
> 19C的数据表同步到Doris中，在初始化快照阶段，同步了一部分数据但还没有到增量阶段，此时取消CDC任务并保存Flink
> Savepoint,取消任务失败；而在任务进入增量阶段后，取消任务并保存savepoint是可以的，请问存量数据同步阶段，为何savepoint失败？
> !image-2024-04-07-17-35-23-136.png!
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
