hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "stack (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-10569) Co-locate meta and master
Date Wed, 14 May 2014 03:55:15 GMT

    [ https://issues.apache.org/jira/browse/HBASE-10569?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13997248#comment-13997248

stack commented on HBASE-10569:

bq. ...what is the benefit of having this complexity.

Complex because we need to be able to support both topologies? (The patch is taking us toward
a simpler deploy and a master state that is easier to reason about; these is also some nice
refactoring of the megalithic  master and regionserver classes).

I could buy the argument that for 1.0 we should just go w/ the old topology and do clean switch
to new layout in 2.0.  I would like it to be on sooner than this but we are short on testing
as it is so one less option, is likely the way to go.

bq. We should not do this for saving on RPC's, but actually getting rid of master in-memory

This is not about saving RPCs.  It is not about getting rid of master-in-memory state either.
 If anything, it is about more of the cluster state being owned by master.

bq. If we have the state in meta and only in meta, then we won't need colocation requirements
or meta-as-a-single-region.

Not sure how this would work.  meta is dumb.  You need an agent of some kind.   I'd like to
hear more E.

This issue has gotten a bit messy ([~toffer] called it first in attached doc saying we seem
to be talking about what this issue is not about).  It is raising loads of good stuff that
we need to break out and make sure it all gets covered (I opened HBASE-11165 at Francis prompting).
 I could give this a pass and add links to our 1.0 issue.

> Co-locate meta and master
> -------------------------
>                 Key: HBASE-10569
>                 URL: https://issues.apache.org/jira/browse/HBASE-10569
>             Project: HBase
>          Issue Type: Improvement
>          Components: master, Region Assignment
>            Reporter: Jimmy Xiang
>            Assignee: Jimmy Xiang
>             Fix For: 0.99.0
>         Attachments: Co-locateMetaAndMasterHBASE-10569.pdf, hbase-10569_v1.patch, hbase-10569_v2.patch,
hbase-10569_v3.1.patch, hbase-10569_v3.patch, master_rs.pdf
> I was thinking simplifying/improving the region assignments. The first step is to co-locate
the meta and the master as many people agreed on HBASE-5487.

This message was sent by Atlassian JIRA

View raw message