You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hudi.apache.org by "Danny Chen (Jira)" <ji...@apache.org> on 2022/05/20 06:03:00 UTC

[jira] [Updated] (HUDI-4129) No need to refresh the timeline server explicitly for RemoteHoodieTableFileSystemView#refresh

     [ https://issues.apache.org/jira/browse/HUDI-4129?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Danny Chen updated HUDI-4129:
-----------------------------
    Description: 
Currently, the {{RequestHandler}} already handles the #sync actively when it finds the server local timeline is behind the client timeline, so we can only refresh the client timeline for {{RemoteHoodieTableFileSystemView}} for two benefits:

1. To avoid to refresh the timeline service too frequently especially for big table, the server would be busy and there is risk for response error, after this change, the refresh only happens when needed
2. For the same instant, we only need to refresh the server once for multiple client requests, compared to current redundant refresh 

> No need to refresh the timeline server explicitly for RemoteHoodieTableFileSystemView#refresh
> ---------------------------------------------------------------------------------------------
>
>                 Key: HUDI-4129
>                 URL: https://issues.apache.org/jira/browse/HUDI-4129
>             Project: Apache Hudi
>          Issue Type: Improvement
>          Components: core
>            Reporter: Danny Chen
>            Priority: Major
>             Fix For: 0.11.1, 0.12.0
>
>
> Currently, the {{RequestHandler}} already handles the #sync actively when it finds the server local timeline is behind the client timeline, so we can only refresh the client timeline for {{RemoteHoodieTableFileSystemView}} for two benefits:
> 1. To avoid to refresh the timeline service too frequently especially for big table, the server would be busy and there is risk for response error, after this change, the refresh only happens when needed
> 2. For the same instant, we only need to refresh the server once for multiple client requests, compared to current redundant refresh 



--
This message was sent by Atlassian Jira
(v8.20.7#820007)