Hi, Martijn,
Thanks for your information. It seems that the situation is similar to
what I know, I will follow FLIP-190.
Also congratulations on becoming a Flink committer!
Best regards,
Chen Zihao
Martijn Visser <martijnvis...@apache.org> 于2022年3月4日周五 16:18写道：
> Hi,
>
> Per the documentation [1] stateful upgrades for SQL are currently not
> supported when upgrading from one minor version to another. There's ongoing
> work to improve this (via FLIP-190 [2]) but that's currently not yet
> available.
>
> Best regards,
>
> Martijn Visser
> https://twitter.com/MartijnVisser82
>
> [1]
> https://nightlies.apache.org/flink/flink-docs-master/docs/dev/table/concepts/overview/#stateful-upgrades-and-evolution
> [2]
> https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=191336489
>
>
> On Fri, 4 Mar 2022 at 04:23, zihao chen <ischenzi...@gmail.com> wrote:
>
>> hi, All.
>>
>> I would like to upgrade the Flink version of several FlinkSQLs from 1.10
>> to 1.12. And I want to restore the state saved by version 1.10 on version
>> 1.12.
>>
>> After I look at the StreamGraph, JobGraph, Checkpoint and some other
>> related informations, I found these points that would cause the state can
>> not be restored.
>> 1. The optimized logical plan has changed in some cases.
>> 2. The structure of JobGraph has changed, and therefore the JobVertexID
>> has changed.
>>
>> During the checkpoint recovery phase, JobManager looks for state through
>> the subtaskID and then performs compatibility checks. Therefore, I feel
>> that to restore the checkpoint, either to ensure that the structure is
>> unchanged, or to find the state of corresponding stateful operators.
>>
>> Do you have any suggestions or experiences with state recovery during
>> version upgrade?
>>
>
