Mailing-List: contact issues-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Date: Wed, 9 Dec 2015 08:24:10 +0000 (UTC)
From: "Elliott Clark (JIRA)" <jira@apache.org>
To: issues@hbase.apache.org
Message-ID: <JIRA.12919905.1449535353000.310320.1449649450975@Atlassian.JIRA>
In-Reply-To: <JIRA.12919905.1449535353000@Atlassian.JIRA>
References: <JIRA.12919905.1449535353000@Atlassian.JIRA>
 <JIRA.12919905.1449535353208@arcas>
Subject: [jira] [Commented] (HBASE-14946) Don't allow multi's to over run
 the max result size.
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HBASE-14946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15048276#comment-15048276 ] 

Elliott Clark commented on HBASE-14946:
---------------------------------------

bq.It is not important that it be accurate?
Not 100% accurate. Just making sure to get an estimate of the size. If we're off by a byte here or there it's not a big deal.

bq.Volatile? Or it don't matter? Or one thread only?
One thread only. 

bq.throw new HBaseIOException("Response size would be too large");
Can do.

bq.So, we are going to break the client response? How do they get the full response back? Needs admin intervention?
The async process should retry all failed gets. Let me get a test to show that.

bq.Why does the scanner chunking not help here?
Multi actions won't contain scans. And we don't chunk on anything where is isGetScan is true.

> Don't allow multi's to over run the max result size.
> ----------------------------------------------------
>
>                 Key: HBASE-14946
>                 URL: https://issues.apache.org/jira/browse/HBASE-14946
>             Project: HBase
>          Issue Type: Improvement
>    Affects Versions: 2.0.0, 1.2.0, 1.3.0
>            Reporter: Elliott Clark
>            Assignee: Elliott Clark
>            Priority: Critical
>         Attachments: HBASE-14946-v1.patch, HBASE-14946-v2.patch, HBASE-14946-v3.patch, HBASE-14946-v5.patch, HBASE-14946.patch
>
>
> If a user puts a list of tons of different gets into a table we will then send them along in a multi. The server un-wraps each get in the multi. While no single get may be over the size limit the total might be.
> We should protect the server from this. 
> We should batch up on the server side so each RPC is smaller.


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