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-15082) Fix merge of MVCC and SequenceID performance regression
Date Mon, 11 Jan 2016 17:43:39 GMT

    [ https://issues.apache.org/jira/browse/HBASE-15082?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15092358#comment-15092358

stack commented on HBASE-15082:

bq. But this is a big change, so I'd say 1.3 is better.

Fine. Could have a 1.3 that comes quickly after 1.2. Means 1.2 has slow increment/append.
 Would have to write this up in known issues.

[~ram_krish] and [~anoop.hbase] thanks for taking a look at this big patch. Was going to post
to rb but wanted to get a unit test run in first to see if basically works. There is a last
piece outstanding -- a read uncommitted when doing the get for an increment/checkAndPut, etc.
(was going to check w/ [~lhofhansl]) -- that I need to fill in still.

Let me do the tag iterator you suggest, the isolation propagation, and the nice find on mvccNum
and getReadpoint.... 

bq. Completed the mvcc and so fwded the readPnt ...So we no longer need persist the seqId
in Cells?

I'd add a check after flush takes update write lock and after it gets its flush sequenceid
to see that mvcc readpt == writept. It should be the case as per your reasoning. If so, yeah,
there should not be a case where an ongoing scanner is going against a read point that could
span memstore and hfile -- so no need of sequenceid to come in when a read from hfile.  Lets

I need to write up our new regime; the one that started at hbase-12751 with its read/write
locking on rows.

> Fix merge of MVCC and SequenceID performance regression
> -------------------------------------------------------
>                 Key: HBASE-15082
>                 URL: https://issues.apache.org/jira/browse/HBASE-15082
>             Project: HBase
>          Issue Type: Sub-task
>          Components: Performance
>            Reporter: stack
>            Assignee: stack
>            Priority: Critical
>             Fix For: 2.0.0, 1.2.0, 1.3.0
>         Attachments: 15082.patch, 15082v2.patch, 15082v2.txt, 15082v3.txt, 15082v4.patch,
15082v5.patch, 15082v6.patch, 15082v7.patch
> This is general fix for increments (appends, checkAnd*) perf-regression identified in
the parent issue. HBASE-15031 has a narrow fix for branch-1.1 and branch-1.0.

This message was sent by Atlassian JIRA

View raw message