Mailing-List: contact dev-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@hbase.apache.org
Received-SPF: pass (nike.apache.org: domain of haosdent@gmail.com designates
 209.85.160.43 as permitted sender)
MIME-Version: 1.0
In-Reply-To: 
 <CAFLnt_o3kL3xrvdo5GNHzaC3mKQAe=4iq-DTSwy3p_CspqZiag@mail.gmail.com>
References: 
 <CAFLnt_qAkDSwdE3=2QMHzPK2shvwAbcdYGhJzqDd0xfNF=u0fg@mail.gmail.com>
	<CAFt=RONOM_7Jm+tGTSoZeTy-SkuqRzTfA40a9NKv6qNpTH=Sbg@mail.gmail.com>
	<CAFLnt_o3kL3xrvdo5GNHzaC3mKQAe=4iq-DTSwy3p_CspqZiag@mail.gmail.com>
Date: Fri, 8 Nov 2013 22:57:40 +0800
Message-ID: 
 <CAFt=ROOMpX1VXGRQtcbwh6j0k8AWD+G1SdEBnxG9ygpyhnKaKg@mail.gmail.com>
Subject: Re: Calling o/s.flush() in HLog.sync()?
From: Haosong Huang <haosdent@gmail.com>
To: dev@hbase.apache.org
Content-Type: multipart/alternative; boundary=001a113630e092082704eaab9a69

--001a113630e092082704eaab9a69
Content-Type: text/plain; charset=GB2312
Content-Transfer-Encoding: quoted-printable

Sorry, I misunderstood what you ask.
2013-11-6 =C9=CF=CE=E79:23=D3=DA "Himanshu Vashishtha" <hv.csuoa@gmail.com>=
=D0=B4=B5=C0=A3=BA

> Okay, good to know but not sure how your response is related to what I
> asked.
>
>
> On Tue, Nov 5, 2013 at 1:08 AM, Haosong Huang <haosdent@gmail.com> wrote:
>
> > An os fsync() call will spent nearly 10ms because of the harddisk iops
> > neckbottle. A hsync() would become two os fsync(). One for checksum fil=
e
> > and theother for block file. If you use SSD disk, you could try use
> fsync()
> > instead of flush() and mount file system without writebarrier.
> > 2013-11-5 =CF=C2=CE=E71:12=D3=DA "Himanshu Vashishtha" <hv.csuoa@gmail.=
com>=D0=B4=B5=C0=A3=BA
> >
> > > Looking at ProtobufLogWriter class, it looks like the call to flush()
> in
> > > the sync method is a noop.
> > >
> > >
> > >
> >
> https://github.com/apache/hbase/blob/trunk/hbase-server/src/main/java/org=
/apache/hadoop/hbase/regionserver/wal/ProtobufLogWriter.java#L134
> > >
> > > The underlying output stream is DFSOutputStream, which doesn't
> implement
> > > flush().
> > >
> > > And, it calls sync() anyway, which ensures the data is written to DN'=
s
> > > (cache).
> > >
> > > Previously with SequenceFile$Writer, it writes data to the outputstre=
am
> > > (using Writables#write), and invoke sync/hflush.
> > >
> > >
> >
> https://github.com/apache/hadoop-common/blob/trunk/hadoop-common-project/=
hadoop-common/src/main/java/org/apache/hadoop/io/SequenceFile.java#L1314
> > >
> > > Is there a reason we have this call here? Please let me know if I mis=
s
> > any
> > > context.
> > >
> > > Thanks,
> > > Himanshu
> > >
> >
>

--001a113630e092082704eaab9a69--