hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "HBase Review Board (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HBASE-2553) Revisit IncrementColumnValue implementation in 0.22
Date Fri, 16 Jul 2010 23:17:52 GMT

    [ https://issues.apache.org/jira/browse/HBASE-2553?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12889398#action_12889398
] 

HBase Review Board commented on HBASE-2553:
-------------------------------------------

Message from: "Ryan Rawson" <ryanobjc@gmail.com>


bq.  On 2010-07-16 13:23:13, stack wrote:
bq.  > src/main/java/org/apache/hadoop/hbase/regionserver/MemStore.java, line 404
bq.  > <http://review.hbase.org/r/325/diff/1/?file=2838#file2838line404>
bq.  >
bq.  >     See KeyValue.matchingQualifier() or KeyValue.matchingColumn

they dont take offset,length so i wasnt able to use them


bq.  On 2010-07-16 13:23:13, stack wrote:
bq.  > src/main/java/org/apache/hadoop/hbase/regionserver/MemStore.java, line 386
bq.  > <http://review.hbase.org/r/325/diff/1/?file=2838#file2838line386>
bq.  >
bq.  >     Why not just use newKv?  If there is a kv in front of newKv, then our counters
are going to be off?

good point


bq.  On 2010-07-16 13:23:13, stack wrote:
bq.  > src/main/java/org/apache/hadoop/hbase/regionserver/Store.java, line 1301
bq.  > <http://review.hbase.org/r/325/diff/1/?file=2839#file2839line1301>
bq.  >
bq.  >     There is a tab here?

fixed


- Ryan


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://review.hbase.org/r/325/#review412
-----------------------------------------------------------





> Revisit IncrementColumnValue implementation in 0.22
> ---------------------------------------------------
>
>                 Key: HBASE-2553
>                 URL: https://issues.apache.org/jira/browse/HBASE-2553
>             Project: HBase
>          Issue Type: Bug
>            Reporter: ryan rawson
>            Assignee: ryan rawson
>             Fix For: 0.90.0
>
>
> right now we are using too much of the old get code, we need to review that and constrain
how this works but without breaking ICV.
> Also we should be resetting the timestamp on every ICV call, and removing the older version.
 Instead of 'updating' an ICV "in place" we should be adding a new one, removing the old one
from memstore (if it is there).  This will play well with the atomic approach added in HBASE-2248
since we are only touching 1 KeyValue at a time.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message