You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Bo Cui (Jira)" <ji...@apache.org> on 2020/08/29 10:51:00 UTC
[jira] [Comment Edited] (HBASE-24925) SCP reduce unnecessary get
requests
[ https://issues.apache.org/jira/browse/HBASE-24925?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17186924#comment-17186924 ]
Bo Cui edited comment on HBASE-24925 at 8/29/20, 10:50 AM:
-----------------------------------------------------------
!image-2020-08-29-17-46-00-900.png!
If the thread pool is not used, load 10k tablestate needs 170+s
was (Author: bo cui):
!image-2020-08-29-17-46-00-900.png!
If the thread pool is not used, load tablestate needs 170+s
> SCP reduce unnecessary get requests
> -----------------------------------
>
> Key: HBASE-24925
> URL: https://issues.apache.org/jira/browse/HBASE-24925
> Project: HBase
> Issue Type: Improvement
> Components: MTTR
> Affects Versions: 3.0.0-alpha-1, 2.2.3
> Reporter: Bo Cui
> Assignee: Bo Cui
> Priority: Major
> Attachments: image-2020-08-29-17-46-00-900.png
>
>
> SCP should reduce unnecessary Get request
> [https://github.com/apache/hbase/blob/65d28da7c22382e040363c607840d5ab6e6b45da/hbase-server/src/main/java/org/apache/hadoop/hbase/master/procedure/ServerCrashProcedure.java#L520]
> during startup， the tableNam2State of tableStateManager is not loading tableState data form metaTable yet. if procThread num is 50 and hbase has 10K tables, in the worst case, the master needs to query meta table 500K times(50*10K. and the regions that all SCPs simultaneously check tableState belong to the same table )
>
> i think master can reduce Get request, and AM#loadMeta can load regions and all tables through asynchronous threads.
> [https://github.com/apache/hbase/blob/65d28da7c22382e040363c607840d5ab6e6b45da/hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/AssignmentManager.java#L1532]
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)