From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jes Sorensen Subject: Re: [PATCH v2 0/4] add --preboot argument to mdadm/mdmon Date: Mon, 23 Jan 2012 19:05:32 +0100 Message-ID: <4F1DA16C.7030202@redhat.com> References: <1327317505-14782-1-git-send-email-Jes.Sorensen@redhat.com> <4F1D89B9.9080802@redhat.com> <4F1D8D27.5070207@redhat.com> <4F1D905C.1050300@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4F1D905C.1050300@redhat.com> Sender: linux-raid-owner@vger.kernel.org To: Doug Ledford Cc: neilb@suse.de, linux-raid@vger.kernel.org, kay@redhat.com, harald@redhat.com, lpoetter@redhat.com, mschmidt@redhat.com List-Id: linux-raid.ids On 01/23/12 17:52, Doug Ledford wrote: >> Kay suggested --preroot on irc - would that do the job? I think that is >> > a bit better than --offroot. > I'd say preroot is better than preboot for sure. I think I still like > offroot more than preroot though. We *are* running off root (from the > initramfs, and we are modifying our argv[0] to signal to systemd that we > are doing exactly this). We *are* keeping our files off of root (in > /run/mdadm versus /var/run/mdadm). And we are intending to live both > before root (preroot) and also after root (postroot). So, I find > preroot to be better than preboot, but still not as accurate as offroot. > Offroot fairly accurately captures what we are doing (although it > doesn't speak to the expected life span of the action, nor does it speak > to the fact that we stay off root only if we are invoked from what will > become the non-root fs, so some sort of check to make sure we are > actually in an initrd/initramfs of sorts might be nice, but I'm not sure > how that would play with things like USB live images where you might > actually want to use the option on a non-initrd environment), where as > both preboot and preroot speak as to when it is used, but not what it > does nor the expected life span of the action. Ok, I guess it depends a bit on how one thinks of it, ie. whether thinking in the context of how it was launched or the system state. I am not really biased one way or another, I mostly care about the functionality, so I am going to wait until tomorrow and see what other input we get and what Neil has to say. Once we have a consensus, I'll cook up a v3 of the patches to match that. Cheers, Jes