[
https://issues.apache.org/jira/browse/FLINK-25322?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17746750#comment-17746750
]
Feifan Wang commented on FLINK-25322:
-------------------------------------
Thanks for reply [~masteryhx] .
{quote}Users have to check the status of restore mode by the Flink UI or the
REST API, right ?
{quote}
Yes, if users want to delete the restored non claimed checkpoint, he/she must
check the flink ui or REST API to confirm the new job is state self-sustained.
Otherwise, new jobs that are not state self-sustaining may fail to restart
because of file not found. But I don't think this adds complexity to the user,
because before that the user also has to check the Flink UI or REST UI to
determine whether the new job has completed at least one checkpoint. In
contrast, I think checking whether the job is state self-sustained is more
intuitive.
{quote}If users stop the job before the 'slowest materilization of all
subtasks', this behaves like LEGACY mode, otherwise this could behaves like
NO_CLIAM mode, right ?
{quote}
In fact, I am also thinking about how to deal with retained checkpoints before
the job reaches state self-sustained. One option is that checkpoints before
state self-sustained will not be retained as retained checkpoints. But this
will cause data duplication when the job using the transactional sink resumes
from that restored checkpoint. Another option is to record in the checkpoint
metadata which state artifacts are borrowed from the non-claimed checkpoint,
and when the new checkpoint is used for claim mode recovery, those state
artifacts borrowed from the non-claimed checkpoint will not be deleted. Do you
have any thoughts on this issue [~pnowojski] ?
{quote}Of course, IIUC, If users want to use NO_CLAIM mode, they'd like to
retain a CP to let other jobs use.
{quote}
In fact, even if a user manually redeploys the same job after updating the
business logic code, it is desirable to be able to use the no-claim mode.
Because the no-claim mode can guarantee that the job can be rolled back when
there is a problem with the new logic code.
> Support no-claim mode in changelog state backend
> ------------------------------------------------
>
> Key: FLINK-25322
> URL: https://issues.apache.org/jira/browse/FLINK-25322
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Checkpointing, Runtime / State Backends
> Reporter: Dawid Wysakowicz
> Assignee: Feifan Wang
> Priority: Major
> Fix For: 1.18.0
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
