[systemd-devel] /usr on separate file system
Tomasz Torcz
tomek at pipebreaker.pl
Thu Mar 3 10:21:32 PST 2011
On Thu, Mar 03, 2011 at 06:45:21PM +0100, Lennart Poettering wrote:
> On Thu, 03.03.11 17:58, Tollef Fog Heen (tfheen at err.no) wrote:
>
> >
> > ]] Lennart Poettering
> >
> > | > To boot a system, enough must be present on the root partition to
> > | > mount other filesystems. This includes utilities, configuration,
> > | > boot loader information, and other essential start-up data. /usr,
> > | > /opt, and /var are designed such that they may be located on other
> > | > partitions or filesystems.
> > |
> > | Well, turns out no distro really follows the spec here, do they?
> >
> > Given the number of Debian people I see running with separate /usr, I
> > believe it works just fine there and while it's a supported
> > configuration I'll patch the warning out of the Debian systemd packages,
> > at least.
>
> Well, it's of course up to you guys what you do there.
>
> But it's a promise you are making there that you cannot keep. If you
> want to support /usr on a separate partition then you'd need to do all
> the work and move the PCI and USB databases to /, move libatasmart,
> fix udisks, fix D-Bus and so on.
> The fact that most these things fail relatively gracefully should not
> mislead you to believe that everything worked fine. Things still fail,
> just not in a big gigantic atomic explosion scenario.
I don't get it. What during the boot (before /usr is mounted) require pci.db,
usb ids, why udisks would be started? I understand that full desktop session
need access to those, but we are talking about short window before starting
system and mounting /usr. What will break? Are the some udev rules needing
mapping between PCI ID and a name? Anything else?
(BTW, “yum remove libatasmart” suggest removal of udisks, nautilus, gvfs,
evolution and some GNOME parts. Nothing related to boot).
--
Tomasz Torcz 72->| 80->|
xmpp: zdzichubg at chrome.pl 72->| 80->|
More information about the systemd-devel
mailing list