accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "John Vines (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-901) Decomissioning servers should focus solely on minor compactions
Date Wed, 12 Dec 2012 20:30:19 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-901?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13530294#comment-13530294
] 

John Vines commented on ACCUMULO-901:
-------------------------------------

Does the master unload them all simultaneously or does it do it serially? If the former, then
they shouldn't try to major compact before unloading. If the latter, perhaps we should introduce
a majC halt as well.
                
> Decomissioning servers should focus solely on minor compactions
> ---------------------------------------------------------------
>
>                 Key: ACCUMULO-901
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-901
>             Project: Accumulo
>          Issue Type: Improvement
>          Components: tserver
>    Affects Versions: 1.4.1
>            Reporter: John Vines
>            Assignee: Keith Turner
>
> I am working on a cluster that has limited resources, which results in major backlogs
of minor and major compactions. I wanted to decomission one to do some disk benchmarking,
which takes a significant amount of time. One of the things I've noticed is that it seems
to keep major compactions going as well.
> I think that when a tserver gets stopped it should drop all queued major compactions,
and block future ones from schedules to A. devote more resources to going through minor compactions
and B. encourage major compaction on the new tserver to induce locality.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message