Return-Path: Delivered-To: apmail-cassandra-user-archive@www.apache.org Received: (qmail 25628 invoked from network); 2 Mar 2011 22:17:03 -0000 Received: from hermes.apache.org (HELO mail.apache.org) (140.211.11.3) by minotaur.apache.org with SMTP; 2 Mar 2011 22:17:03 -0000 Received: (qmail 47867 invoked by uid 500); 2 Mar 2011 22:17:00 -0000 Delivered-To: apmail-cassandra-user-archive@cassandra.apache.org Received: (qmail 47829 invoked by uid 500); 2 Mar 2011 22:17:00 -0000 Mailing-List: contact user-help@cassandra.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: user@cassandra.apache.org Delivered-To: mailing list user@cassandra.apache.org Received: (qmail 47821 invoked by uid 99); 2 Mar 2011 22:17:00 -0000 Received: from athena.apache.org (HELO athena.apache.org) (140.211.11.136) by apache.org (qpsmtpd/0.29) with ESMTP; Wed, 02 Mar 2011 22:17:00 +0000 X-ASF-Spam-Status: No, hits=2.9 required=5.0 tests=HTML_MESSAGE,MIME_QP_LONG_LINE,RCVD_IN_DNSWL_NONE,SPF_NEUTRAL X-Spam-Check-By: apache.org Received-SPF: neutral (athena.apache.org: local policy) Received: from [66.33.216.122] (HELO hapkido.dreamhost.com) (66.33.216.122) by apache.org (qpsmtpd/0.29) with ESMTP; Wed, 02 Mar 2011 22:16:55 +0000 Received: from homiemail-a55.g.dreamhost.com (caibbdcaaaaf.dreamhost.com [208.113.200.5]) by hapkido.dreamhost.com (Postfix) with ESMTP id AC1D2178F60 for ; Wed, 2 Mar 2011 14:16:34 -0800 (PST) Received: from homiemail-a55.g.dreamhost.com (localhost [127.0.0.1]) by homiemail-a55.g.dreamhost.com (Postfix) with ESMTP id 4552112C054 for ; Wed, 2 Mar 2011 14:16:28 -0800 (PST) DomainKey-Signature: a=rsa-sha1; c=nofws; d=thelastpickle.com; h=to:from :subject:date:message-id:content-type:mime-version:in-reply-to; q=dns; s=thelastpickle.com; b=N5uasTLfLqozg5+RUj0lVAZrTl+nRyp+z zc0F6QBdyfKBEA+dmyZ830bEwddRKYh0mnyR7Edxpzbc8Rg0Jt7ISEzWIDQpB3dj 0Y/OSrir16uZAWfDNNxBo2BLn/AcY4JWQhOcMDhklIawPs3k8W1iD6lauEsJSDLB F0jhAOrfho= DKIM-Signature: v=1; a=rsa-sha1; c=relaxed; d=thelastpickle.com; h=to :from:subject:date:message-id:content-type:mime-version: in-reply-to; s=thelastpickle.com; bh=odEhsWYFjWXRi+1QMKZvOaqzOLg =; b=flWYvbXyv7EdQCXZ9aaXu+000GB/fxSVCqCIyAVql70hY1Eis5W1eZ+37mn jXT6Rtub+bHpGDghXtBH7HrY6TFgaTR8iwYBUv9F2vaIKkDhuKvED4U3Z5/nPLdJ lBEbeQk2I9O82hVTLGqCP9KrJ86Ajg8fbAJ2YyJgFhJ7+7GI= Received: from localhost (webms.mac.com [17.148.16.116]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) (Authenticated sender: aaron@thelastpickle.com) by homiemail-a55.g.dreamhost.com (Postfix) with ESMTPSA id 21B2312C03C for ; Wed, 2 Mar 2011 14:16:28 -0800 (PST) To: user@cassandra.apache.org From: Aaron Morton Subject: Re: Seed Nodes Date: Wed, 02 Mar 2011 22:16:27 GMT X-Mailer: MobileMe Mail (1C3224) Message-id: <9edd2f53-3512-4339-9f80-896255769a96@me.com> Content-Type: multipart/alternative; boundary=Apple-Webmail-42--02c4bf47-dbc0-4ceb-8127-d4db43f57fa3 MIME-Version: 1.0 In-Reply-To: --Apple-Webmail-42--02c4bf47-dbc0-4ceb-8127-d4db43f57fa3 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=ISO-8859-1; format=flowed The seed list is also used as part of the Gossip round=A0http://wiki.apach= e.org/cassandra/ArchitectureGossip=A0=0A=0AI agree it's not critical to en= sure all running nodes have the same running seed list. But it's=A0definit= ely=A0a good idea.=A0=0A=0ADisclaimer: Learning more about the gossip prot= ocol is on my to-do list :)=0A=0AAaron=0A=A0=A0=0A=0AOn 03 Mar, 2011,at 08= :27 AM, Nick Bailey wrote:=0A=0AYou do not need to rest= art when changing the seedlist of nodes that are up. The seedlist is reall= y only used when a node first starts. So if you modify the list of seeds t= he changes will take effect whenever that node restarts, exactly when they= need to.=0A=0AIn 0.7 you can basically get away with ignoring the seedlis= t for nodes that have already joined the cluster since the ring state is r= emembered between restarts. "Best Practices" may dictate that you keep all= your seed lists updated though.=0A=0A=0AOn Tue, Mar 1, 2011 at 9:29 PM, E= ric Gilmore wrote:=0AYes, two per DC is a recommendati= on I've heard from Jonathan Ellis.=A0 We put that in yet more documentatio= n at http://www.datastax.com/dev/tutorials/getting_started_0.7/configuring= #seed-list (appreciate the citation Aaron :)=0A=0AI had a recent conversat= ion with a Cassandra expert who had me convinced that, as long as a node o= nline already had one seed in its list at startup, you wouldn't really nee= d to restart it -- at least right away -- after adding a second seed to th= e config.=A0 Rather than confuse the issue by trying to remember/recreate = the rationale for that, I'll ping him and see if he'll comment here.=0A=0A= =0A=0AOn Tue, Mar 1, 2011 at 12:04 PM, Aaron Morton wrote:=0AAFAIK it's recommended to have two seed nodes per dc.=0A=0ASo= me info on seeds here=A0http://www.datastax.com/docs/0.7/operations/cluste= ring=0A=0AYou will need a restart.=0A=0AAaron=0A=0A=0A=0AOn 2/03/2011, at = 6:08 AM, shan.lu@accenture.com wrote:=0A=0A=0A=0AHow many seed nodes shoul= d I have for a cluster of 100 nodes each with about 500gb of data? Also to= add seeds the nodes, must I change the seed nodes list on all existing no= des through the Cassandra.yaml file? Will changes take effect without rest= arting the node?=0A=A0=0AShan (Susie) Lu,=A0 Analyst=0AAccenture Technolog= y Labs - Silicon Valley=0Acell=A0 +1 425.749.2546=A0=A0=A0=0Aemail=A0shan.= lu@accenture.com=0A=A0=0A=0AThis message is for the designated recipient o= nly and may contain privileged, proprietary, or otherwise private informat= ion. If you have received it in error, please notify the sender immediatel= y and delete the original. Any other use of the email by you is prohibited= =0A=0A=0A --Apple-Webmail-42--02c4bf47-dbc0-4ceb-8127-d4db43f57fa3 Content-Type: multipart/related; type="text/html"; boundary=Apple-Webmail-86--02c4bf47-dbc0-4ceb-8127-d4db43f57fa3 --Apple-Webmail-86--02c4bf47-dbc0-4ceb-8127-d4db43f57fa3 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=ISO-8859-1;
The seed list is also used as part of the Gossip round http://wiki.apache.o= rg/cassandra/ArchitectureGossip 

I agree it's = not critical to ensure all running nodes have the same running seed list. = But it's definitely a good idea. 

Disclai= mer: Learning more about the gossip protocol is on my to-do list :)
<= div>
Aaron
  

On 03 Mar, 2= 011,at 08:27 AM, Nick Bailey <nick@riptano.com> wrote:

=
You do not need to restart when changing the s= eedlist of nodes that are up. The seedlist is really only used when a node= first starts. So if you modify the list of seeds the changes will take ef= fect whenever that node restarts, exactly when they need to.
=0A=0A=
In 0.7 you can basically get away with ignoring the seedlist fo= r nodes that have already joined the cluster since the ring state is remem= bered between restarts. "Best Practices" may dictate that you keep all you= r seed lists updated though.


On Tue, Mar 1, 2011 at 9:29 PM, Eric Gilmore <<= a href=3D"mailto:eric@datastax.com" _mce_href=3D"mailto:eric@datastax.com"= >eric@datastax.com> wrote:
=0A=0AYes, two per DC is a recommendation I've heard from Jona= than Ellis.  We put that in yet more documentation at http://www.datastax.com/dev/tutorials/getting_start= ed_0.7/configuring#seed-list (appreciate the citation Aaron :)
=0A=0A= =0A
I had a recent conversation with a Cassandra expert who had me conv= inced that, as long as a node online already had one seed in its list at s= tartup, you wouldn't really need to restart it -- at least right away -- a= fter adding a second seed to the config.  Rather than confuse the iss= ue by trying to remember/recreate the rationale for that, I'll ping him an= d see if he'll comment here.
=0A=0A

=0A
On Tue, Mar 1, 2011 at 12:04 PM, Aaron= Morton <aaron@thelastpickle.com>= ; wrote:
=0A=0A=0A
AFAIK it's recommended to have two see= d nodes per dc.

=0A=0A=0A

You wi= ll need a restart.

Aaron<= /font>



On 2/03/2011, at 6:08 AM, = shan.lu@accenture.com wrote:
=0A=0A=0A


=0A
=0A=

How many seed nodes should I have for a cluster of = 100 nodes each with about 500gb of data? Also to add seeds the nodes, must= I change the seed nodes list on all existing nodes through the Cassandra.= yaml file? Will changes take effect without=0A restarting the node?

=0A=

 

=0A

Shan (Susie) Lu, =0AAnalyst<= /b>

=0A

Accenture Technology = Labs - Silicon Valley=0A

=0A

cell  +1 425.749.2546  &nbs= p;

=0A

email shan.lu@accenture.com<= /span>

=0A=0A=0A=0A

 

=0A
=0A
=0A=
=0AThis message is for = the designated recipient only and may contain privileged, proprietary, or = otherwise private information. If you have received it in error, please no= tify the sender immediately and delete the original.=0A Any other use of t= he email by you is prohibited.
=0A=0A=0A=0A

=0A
<= br>=0A
--Apple-Webmail-86--02c4bf47-dbc0-4ceb-8127-d4db43f57fa3-- --Apple-Webmail-42--02c4bf47-dbc0-4ceb-8127-d4db43f57fa3--