hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sean Busbey (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-12483) Refactor WAL recovery to abstract away filesystem
Date Thu, 04 Dec 2014 03:59:12 GMT

    [ https://issues.apache.org/jira/browse/HBASE-12483?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14233889#comment-14233889
] 

Sean Busbey commented on HBASE-12483:
-------------------------------------

nothing yet. I've got my hands full  this week with HBASE-11983 and HBASE-5699.

If you have time and want to start working on it, let me know. Happy to do reviews.

> Refactor WAL recovery to abstract away filesystem
> -------------------------------------------------
>
>                 Key: HBASE-12483
>                 URL: https://issues.apache.org/jira/browse/HBASE-12483
>             Project: HBase
>          Issue Type: Improvement
>          Components: wal
>            Reporter: Sean Busbey
>            Assignee: Sean Busbey
>
> HBASE-10378 hides the filesystem from the write-to-the-wal path, but leaves in place
the original recovery code. Refactor to remove this leak.
> * Master should have means to say "regions from these servers need recovery"
> * Splitting work should be phrased in terms of region groups and regions from a server
instead of files.
> * RegionServers doing recovery should be able to say "please replay the edits from this
region"
> * Similarly Replication should be asking for replay of edits as data is aged out of the
wal rather than looking specifically at file archiving.



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

Mime
View raw message