Mailing-List: contact user-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@cassandra.apache.org
Received-SPF: pass (athena.apache.org: domain of iecanfly@gmail.com designates
 74.125.82.172 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <F82673AF-CFE4-4D13-B84C-100BD88888C7@gmail.com>
References: 
 <CANyEDMDdDScTxXGKSBKSwKH2e+qEOJq1krfUvgZ3GpZefpfF2g@mail.gmail.com>
	<F82673AF-CFE4-4D13-B84C-100BD88888C7@gmail.com>
Date: Thu, 7 Jul 2011 19:57:45 +0900
Message-ID: 
 <CANyEDMAdHTSOUy8sXsxD6Bc7jOGGf4kQPNfrvPRj1U1c_gKjSA@mail.gmail.com>
Subject: Re: minimum number of machines for RF=3
From: david lee <iecanfly@gmail.com>
To: user@cassandra.apache.org
Content-Type: multipart/alternative; boundary=000e0ce0b71a41499804a778965a

--000e0ce0b71a41499804a778965a
Content-Type: text/plain; charset=ISO-8859-1

cheers~

On 7 July 2011 19:41, Watanabe Maki <watanabe.maki@gmail.com> wrote:

> It is expected behaviour and not relate on number of node.
> After the failed node bringing back, the ring will be busy by Hinted
> Handoff rewriting and Read Repair. If you run repair, all your 3 nodes need
> to build Merkel Tree, compare the hash values,  then transfer latest data to
> each other.
>
> You can tune the HH, read repair to reduce performance impact on self
> healing activities.
>
> maki
>
> On 2011/07/07, at 18:10, david lee <iecanfly@gmail.com> wrote:
>
> > hi guys,
> >
> > is there a minimum(recommended) number of machines for RF=3?
> >
> > i encountered a test result where
> > # of nodes = 3
> > RF  =  3
> > CL.READ=QUORUM
> > and when 1 node was taken out and back in after which node repair was
> run,
> > the TPS dropped significantly.
> >
> > is this behaviour expected since the RF (for the duration of when 1 node
> was taken out)
> > is higher than the # of nodes?
> >
> > cheers,
> > david
>
>

--000e0ce0b71a41499804a778965a
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

cheers~<br><br><div class=3D"gmail_quote">On 7 July 2011 19:41, Watanabe Ma=
ki <span dir=3D"ltr">&lt;<a href=3D"mailto:watanabe.maki@gmail.com">watanab=
e.maki@gmail.com</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote"=
 style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
It is expected behaviour and not relate on number of node.<br>
After the failed node bringing back, the ring will be busy by Hinted Handof=
f rewriting and Read Repair. If you run repair, all your 3 nodes need to bu=
ild Merkel Tree, compare the hash values, =A0then transfer latest data to e=
ach other.<br>

<br>
You can tune the HH, read repair to reduce performance impact on self heali=
ng activities.<br>
<font color=3D"#888888"><br>
maki<br>
</font><div><div></div><div class=3D"h5"><br>
On 2011/07/07, at 18:10, david lee &lt;<a href=3D"mailto:iecanfly@gmail.com=
">iecanfly@gmail.com</a>&gt; wrote:<br>
<br>
&gt; hi guys,<br>
&gt;<br>
&gt; is there a minimum(recommended) number of machines for RF=3D3?<br>
&gt;<br>
&gt; i encountered a test result where<br>
&gt; # of nodes =3D 3<br>
&gt; RF =A0=3D =A03<br>
&gt; CL.READ=3DQUORUM<br>
&gt; and when 1 node was taken out and back in after which node repair was =
run,<br>
&gt; the TPS dropped significantly.<br>
&gt;<br>
&gt; is this behaviour expected since the RF (for the duration of when 1 no=
de was taken out)<br>
&gt; is higher than the # of nodes?<br>
&gt;<br>
&gt; cheers,<br>
&gt; david<br>
<br>
</div></div></blockquote></div><br>

--000e0ce0b71a41499804a778965a--