Mailing-List: contact user-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@hbase.apache.org
Received-SPF: pass (athena.apache.org: domain of saint.ack@gmail.com
 designates 209.85.214.169 as permitted sender)
MIME-Version: 1.0
Sender: saint.ack@gmail.com
In-Reply-To: 
 <CABbGW3wZWg=uL79P90KuSxcCWh05cDM_ygtv60vhOacb3ro9+Q@mail.gmail.com>
References: 
 <CABbGW3wZWg=uL79P90KuSxcCWh05cDM_ygtv60vhOacb3ro9+Q@mail.gmail.com>
Date: Sat, 15 Sep 2012 13:30:54 -0700
Message-ID: 
 <CADcMMgGo5q5rQ=S293g-Btn4=0GP6LuwkzJW3RivccoSbX=TdA@mail.gmail.com>
Subject: Re: support checksums in HBase block
From: Stack <stack@duboce.net>
To: user@hbase.apache.org
Content-Type: text/plain; charset=ISO-8859-1

On Sat, Sep 15, 2012 at 10:31 AM, jlei liu <liulei412@gmail.com> wrote:
> I use Hbase0.94.1 and  hadoop-0.20.2-cdh3u5.
>
> The Hbase0.94.1 write checksums in HBase block, so we don't need to read
> checksums from metadata file of block file. But I find the BlockSender
> still read checksums from metadata file of block file every time,

How are you figuring this?   That sounds like a regression.

In 0.94.x the default is that hbase.regionserver.checksum.verify is
true which should have hbase doing checksum verify on read rather than
hdfs.

>  so the
> Hbase0.94.1 can't decrease the iops, that is right? If that is right,
> should I set the  hbase.regionserver.checksum.verify property to false in
> order to  avoid the Hbase0.94.1 to verify th checksum?
>

It would be interesting flipping the switch.  If things went faster,
then for sure there is a regression.
St.Ack