[ 
https://issues.apache.org/jira/browse/HUDI-4725?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Tan Yajun updated HUDI-4725:
----------------------------
    Description: 
When we use the multi plan compaction,we may need the plan to commit in 
order,so I try to figure that by create a set to store the instant that are not 
ready,and I transform the compactionPlans(List of 
Pair<String(instant),compactionPlan>) to the sinkFunction.When a commit is 
ready,I remove the instant from the noReady cache,and I will check that if it 
the earliest compaction need to commit, if that commit the compaction instant 
and check if the next one is ready and commit.if it's not the earliest one, 
just remove its instant from the noReady cache.

 

[https://github.com/apache/hudi/pull/6512]

  was:
when we use the multi plan,we may need the plan to commit inorder,so I try to 
figure that.

 

https://github.com/apache/hudi/pull/6512


> Add inorder commit for multi compaction plan
> --------------------------------------------
>
>                 Key: HUDI-4725
>                 URL: https://issues.apache.org/jira/browse/HUDI-4725
>             Project: Apache Hudi
>          Issue Type: Improvement
>          Components: compaction
>            Reporter: Tan Yajun
>            Assignee: Tan Yajun
>            Priority: Major
>   Original Estimate: 2h
>  Remaining Estimate: 2h
>
> When we use the multi plan compaction,we may need the plan to commit in 
> order,so I try to figure that by create a set to store the instant that are 
> not ready,and I transform the compactionPlans(List of 
> Pair<String(instant),compactionPlan>) to the sinkFunction.When a commit is 
> ready,I remove the instant from the noReady cache,and I will check that if it 
> the earliest compaction need to commit, if that commit the compaction instant 
> and check if the next one is ready and commit.if it's not the earliest one, 
> just remove its instant from the noReady cache.
>  
> [https://github.com/apache/hudi/pull/6512]



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to