You are viewing a plain text version of this content. The canonical link for it is here.

Posted to commits@hudi.apache.org by "Raymond Xu (Jira)" <ji...@apache.org> on 2022/08/31 06:01:00 UTC

[jira] [Closed] (HUDI-2116) sync 10w partitions to hive by using HiveSyncTool lead to the oom of hive MetaStore

     [ https://issues.apache.org/jira/browse/HUDI-2116?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Raymond Xu closed HUDI-2116.
----------------------------

>  sync 10w partitions to hive by using HiveSyncTool lead to the oom of hive MetaStore
> ------------------------------------------------------------------------------------
>
>                 Key: HUDI-2116
>                 URL: https://issues.apache.org/jira/browse/HUDI-2116
>             Project: Apache Hudi
>          Issue Type: Bug
>          Components: hive
>    Affects Versions: 0.8.0
>         Environment: hive3.1.1
> hadoop 3.1.1
>            Reporter: tao meng
>            Assignee: tao meng
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 0.9.0
>
>
> when we try to sync 10w partitions to hive by using HiveSyncTool lead to the oom of hive MetaStore。
>  
> here is a stress test for HiveSyncTool
> env： 
> hive metastore -Xms16G -Xmx16G
> hive.metastore.client.socket.timeout=10800
>  
> ||partitionNum||time consume||
> |100|37s|
> |1000|168s|
> |5000|1830s|
> |10000|timeout|
> |100000|hive metastore oom|
> HiveSyncTools  sync all partitions to hive metastore at once。 when the partitions num is large ，it puts a lot of pressure on hive metastore。 for large partition num we should support batch sync 。
>  



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