From mboxrd@z Thu Jan 1 00:00:00 1970 From: Neil Brown Subject: Re: mdadm, raid10, and problem Date: Tue, 22 May 2007 10:43:34 +1000 Message-ID: <18002.15542.445123.221297@notabene.brown> References: <20070520153535.GA563@ep09.pld-linux.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: message from Maciej 'Agaran' Pijanka on Sunday May 20 Sender: linux-raid-owner@vger.kernel.org To: Maciej 'Agaran' Pijanka Cc: linux-raid@vger.kernel.org List-Id: linux-raid.ids On Sunday May 20, agaran@pld-linux.org wrote: > > Hello > > > i was trying to build raid 10 from six devices (currently loops but need > to check before i try to do same on physical drives). > > and mdadm gets bit mad, > > mdadm -C -l10 -n6 /dev/loop1 /dev/loop2 /dev/loop3 missing missing > missing > produced weird result What did you expect :-? > mdadm: RUN_ARRAY failed: Input/output error > mdadm: stopped /dev/md0 Yes, lots of mdadm's errors are - like this - quite unhelpful. It is something that I would like to get improved, but it just doesn't get a high priority. > -from dmesg: > md: bind > md: bind > md: bind > md: md0: raid array is not clean -- starting background reconstruction > raid10: not enough operational mirrors for md0 "not enough" means "you don't have the right drives". For every level other than raid10, it doesn't matter which drives you have, it just matter how many there are. For raid10, you do that to have are set of drives that provides all blocks. But I just re-used the error message from raid1. Patches always welcome... NeilBrown