From mboxrd@z Thu Jan 1 00:00:00 1970 From: Joao Eduardo Luis Subject: Re: Documentation Discrepancy - Manual Configuration Date: Wed, 23 Apr 2014 01:23:03 +0100 Message-ID: <535707E7.3070603@inktank.com> References: <6b1a50710afdfd0b10133934be7eac38@pl1.haspere.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from mail-wi0-f173.google.com ([209.85.212.173]:40676 "EHLO mail-wi0-f173.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750852AbaDWAW7 (ORCPT ); Tue, 22 Apr 2014 20:22:59 -0400 Received: by mail-wi0-f173.google.com with SMTP id z2so4111585wiv.12 for ; Tue, 22 Apr 2014 17:22:58 -0700 (PDT) In-Reply-To: <6b1a50710afdfd0b10133934be7eac38@pl1.haspere.com> Sender: ceph-devel-owner@vger.kernel.org List-ID: To: YS3fpFE2ykfB@dyweni.com, ceph-devel@vger.kernel.org On 04/18/2014 08:10 AM, Dyweni - Ceph-Devel wrote: > Hi, > > Testing Ceph with 0.79. > > 1 Monitor > 2 OSDs > > > Following the instructions here > (http://ceph.com/docs/master/install/manual-deployment/)... I arrive at > this state: > > ------ > cluster e3e1a87b-d282-41b5-b4ad-fb3f969e164f > health HEALTH_WARN 192 pgs degraded; 192 pgs stuck unclean > monmap e1: 1 mons at {a=10.208.39.100:6789/0}, election epoch 1, > quorum 0 a > osdmap e16: 2 osds: 2 up, 2 in > pgmap v21: 192 pgs, 3 pools, 0 bytes data, 0 objects > 2004 MB used, 59381 MB / 65536 MB avail > 192 active+degraded Indeed, ceph's default pool size is 3. However, during the manual deployment guide you followed, on the "Monitor Bootstrap" section, 14th step, you certainly configured ceph.conf. During this step, the guide presents you with an example configuration file containing the option 'osd pool default size = 2', which overrides the default pool size of 3. Furthermore, later during the 'Adding OSDs' section, this is made obvious right on the first paragraph. Could this be made more obvious to a new-comer, probably. But the info is there :) -Joao > > 'ceph osd dump' shows this: > ------ > ... > pool 0 'data' replicated size 3 min_size 2 ... > pool 1 'metadata' replicated size 3 min_size 2 ... > pool 2 'rbd' replicated size 3 min_size 2 ... > ... > ------ > > > > It's not until I add a third OSD, or reduce the size of the pools, does > the 192 pgs become unstuck and migrate towards active+clean. > > > > I think the documentation should indicate that three nodes are required > for Ceph to become optimal. > > > -- Joao Eduardo Luis Software Engineer | http://inktank.com | http://ceph.com