'Twas brillig, and Thomas Backlund at 31/10/11 19:01 did gyre and gimble: > Colin Guthrie skrev 31.10.2011 19:47: >>> Just have initramfs mount / and /usr, no need to merge. >> >> Yes, this is exactly what is intended, but it depends on the use case: >> > > Then I have misread something somewhere along the way... :/
Yeah the link Olav provided ( http://www.freedesktop.org/wiki/Software/systemd/separate-usr-is-broken ) is certainly worth reading if you've not already done so. The "Going Forward" section in particular starts with this quote: "/usr on its own filesystem is useful in some custom setups. But instead of expecting the traditional Unix way to (sometimes mindlessly) distributing tools between /usr and /, and require more and more tools to move to /, we now just expect /usr to be pre-mounted from inside the initramfs, to be available before 'init' starts." >> For simple installs, such as on my laptop or typical "desktop" linux, >> then having separate / and /usr is something I've specifically avoided >> for just about ever (I sometimes have separate /boot for "complicated >> rootfs" reasons and I almost always have a separate /home, but other >> than that, I like to keep my desktops/laptops simple). In other words >> there just are not sufficient benefits in doing this. > > Yes, > I agree for the desktop/laptop/pad/ use there is no need for splitting > besides /home and maybe separate /boot if you want to use encryption. Nod >> But on the server, especially on a farm of servers with a potentially >> shared /usr across several machines, then yes, keeping it as something >> that can be mounted by the initramfs is perfectly feasible. This is a >> specific use case that Lennart + Co are trying to design for. >> > > Ok, this is exactly the part that have raised my concerns that it was > about to be "forgotten" in favour of "simplicity". > > Server installs usually want to be more "free" to finetune. > > Another use case is to be able to split up the system on more harddisks > to even the load on the disks. > > And yet another is simply to get more space depending on hw. > (yes I know lvm exists but it's also an extra layer / point of failure > that is not always worth it) Totally agree about LVM (and to a lesser degree the RAID layer)... I'm looking forward to btrfs which merges in the RAID layers and (I think - I've not fully read up on it) the LVM functionality too. Currently, I've found performance really takes a hit with my RAID+LVM setups... 40M/s write speeds where as just plain RAID was getting closer to 160M/s... this is likely in part due to 4k sector sizes too of course. But I digress! > But if this part is the "some specific usecases" then I guess I need to > start re-reading some stuff :) :) Col -- Colin Guthrie colin(at)mageia.org http://colin.guthr.ie/ Day Job: Tribalogic Limited http://www.tribalogic.net/ Open Source: Mageia Contributor http://www.mageia.org/ PulseAudio Hacker http://www.pulseaudio.org/ Trac Hacker http://trac.edgewall.org/