[
https://issues.apache.org/jira/browse/HUDI-1576?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17476506#comment-17476506
]
Nishith Agarwal commented on HUDI-1576:
---------------------------------------
[~guoyihua] Yes, the idea was to detach archiving from being inline to async.
Although, even today, archiving happens after the "COMMIT" is successfully
completed and the file has been created on disk. So, introducing a new action
is not needed. I think archival can just run async and keep archiving contents
without the need to create any action since that may be an overkill. One
side-effect I see is that we still need a way to track the progress and
activity of archiving on a table. Since the .archive folder has this history,
it should be fine. That's my opinion.
> Add ability to perform archival synchronously
> ---------------------------------------------
>
> Key: HUDI-1576
> URL: https://issues.apache.org/jira/browse/HUDI-1576
> Project: Apache Hudi
> Issue Type: Improvement
> Components: Writer Core
> Reporter: Nishith Agarwal
> Assignee: Ethan Guo
> Priority: Blocker
> Fix For: 0.11.0
>
>
> Currently, archival runs inline. We want to move archival to a table service
> like cleaning, compaction etc..
> and treat it like that. of course, no new action will be introduced.
>
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
