From mboxrd@z Thu Jan 1 00:00:00 1970 From: Michael Tokarev Subject: Re: md[adm] device names Date: Mon, 02 Nov 2009 14:21:26 +0300 Message-ID: <4AEEC0B6.8020301@msgid.tls.msk.ru> References: <4AEC33A2.6050408@msgid.tls.msk.ru> <19182.16585.929763.745870@notabene.brown> <20091102082424.GD7626@lapse.rw.madduck.net> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: In-Reply-To: <20091102082424.GD7626@lapse.rw.madduck.net> Sender: linux-raid-owner@vger.kernel.org To: Neil Brown , Michael Tokarev , linux-raid List-Id: linux-raid.ids martin f krafft wrote: > also sprach Neil Brown [2009.11.02.0315 +0100]: >> Numbers are meaningless. I would much rather have "/dev/md/home" or >> "/dev/md/backup" or whatever. But as I said, old names should still >> work. >=20 > =85 except sysfs exposes device names =E0 la md1, right? I agree with > Michael that we should either have numbers or names, but not both. > http://bugs.debian.org/553896 came in today, which is just another > instance of confusion resulting from multiple choices. If I see it right, Neil's plan was to keep /dev/mdN naming intact, _and_ introduce _additional_ /dev/md/friendlyname scheme a-la /dev/disk/by-*/foo. _That_ looks quite good thing. It wasn't an intention to have multiple device nodes. But what bothers me in this whole thing is why, out of the sudden, in initramfs we've /dev/md/0 (in a subdir) instead of /dev/md0 (the traditional way and the kernel device naming). > I personally don't have a problem with numbers; they are as > meaningless or -full as partition numbers, and I used stuff like > /dev/sda5 on my systems for decades without problems. The trend > these days is dm/LVM, and that gives us decent names like > /dev/mapper/home, and I see the benefit in moving towards > /dev/md/home, but it needs to be done consistently, and either one > or the other. No need to _move_ towards /dev/md/home, as long as you're happy with /dev/md1. Having /dev/md1 device _and_ /dev/md/home symlink is handy (And lvm from this point of view is something entirely different since it _never_ had any meaningful numbers). What is questionable is why or how the "main" device node got moved from /dev/md1 to /dev/md/1. Which is what this new bugreport is about. /mjt -- To unsubscribe from this list: send the line "unsubscribe linux-raid" i= n the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html