hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Eli Collins (Commented) (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-5655) Cap space usage of default log4j rolling policy
Date Tue, 27 Mar 2012 22:36:30 GMT

    [ https://issues.apache.org/jira/browse/HBASE-5655?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13240020#comment-13240020
] 

Eli Collins commented on HBASE-5655:
------------------------------------

The issue is that the DRFA doesn't have the ability to cap the the #files or space usage,
so people who aren't using host-level log rotation fill up their log partitions, their clusters
crash hard. The idea here is to change the default to something safe (won't crash your cluster)
and people who want to keep lots of logs can adjust the limit accordingly, eg they can bump
up the default to the size of their log partition. Should make HBase more user-friendly. Given
that the default is 5gb (lots of logs!) I suspect few people will need to do that. 

Ideally we'd be able to keep daily files AND limit space usage, unfortunately the DRFA doesn't
support that =(

Given that this is an incompatible change it may be better to target for 94 or the singularity.
                
> Cap space usage of default log4j rolling policy
> -----------------------------------------------
>
>                 Key: HBASE-5655
>                 URL: https://issues.apache.org/jira/browse/HBASE-5655
>             Project: HBase
>          Issue Type: Improvement
>    Affects Versions: 0.92.1
>            Reporter: Himanshu Vashishtha
>            Assignee: Himanshu Vashishtha
>
> The current default log4j policy is to use Daily Rolling File Appender (DRFA). At times,
its good to have a cap on the maximum size of the logs in order to limit its disk usage. Here
is a proposal to set a new file appemder (RFA) as the default appender. It can be configured
via env so that existing tools can use the current behavior of using DRFA instead. 
> This is in parallel with jira Hadoop-8149.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Mime
View raw message