hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andrew Purtell (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (HBASE-16091) Canary takes lot more time when there are delete markers in the table
Date Fri, 24 Jun 2016 16:15:16 GMT

    [ https://issues.apache.org/jira/browse/HBASE-16091?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15348492#comment-15348492
] 

Andrew Purtell edited comment on HBASE-16091 at 6/24/16 4:14 PM:
-----------------------------------------------------------------

Please don't commit the current version. This behavior should be configurable. As soon as
this behavior can be configurable, preferably on the command line, and defaulting to the current
behavior, then I will withdraw my objection.


was (Author: apurtell):
Please don't commit the current version. This behavior should be configurable.

> Canary takes lot more time when there are delete markers in the table
> ---------------------------------------------------------------------
>
>                 Key: HBASE-16091
>                 URL: https://issues.apache.org/jira/browse/HBASE-16091
>             Project: HBase
>          Issue Type: Bug
>    Affects Versions: 2.0.0
>            Reporter: Vishal Khandelwal
>            Assignee: Vishal Khandelwal
>             Fix For: 1.1.6, 1.2.3, 0.98.22
>
>         Attachments: HBASE-16091.00.patch
>
>
> We have a table which has lot of delete markers and we running Canary test on a regular
interval sometimes tests are timing out because to reading first row would skip all these
delete markers. Since purpose of Canary is to find health of the region, i think keeping raw=true
would not defeat the purpose but provide good perf improvement. 
> Following are the example of one such scan where 
> without changing code it took 62.3 sec for onre region scan
> 2016-06-23 08:49:11,670 INFO  [pool-2-thread-1] tool.Canary - read from region  <tablename>.<region>
column family 0 in 62338ms
> whereas after setting raw=true, it reduced to 58ms
> 2016-06-23 08:45:20,259 INFO  [pool-2-thread-1] tests.Canary - read from region <tablename>.<region>
column family 0 in 58ms
> Taking this over multiple tables , with multiple region would be a good performance gain.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message