Gentoo Archives: gentoo-user

From: Grant Edwards <grant.b.edwards@×××××.com>
To: gentoo-user@l.g.o
Subject: [gentoo-user] Re: Can't find init due to inconsistent drive order
Date: Tue, 02 Jul 2013 02:50:28
Message-Id: kqtf16$pns$2@ger.gmane.org
In Reply to: Re: [gentoo-user] Can't find init due to inconsistent drive order by Paul Hartman
1 On 2013-07-01, Paul Hartman <paul.hartman+gentoo@×××××.com> wrote:
2 > On Mon, Jul 1, 2013 at 4:52 PM, Grant Edwards <grant.b.edwards@×××××.com> wrote:
3 >> I've just recently run into a problem where sometimes when a machine
4 >> boots, the kernel can't find init. This appears to be because my grub
5 >> configuration line says "root=/dev/sda5" and _sometimes_ the drive
6 >> that contains my root partition is sdb instead of sda. AFAICT, for the
7 >> past 30 years the linux kernel was 100% consistent in the order that
8 >> hard drives were labelled -- but recently that has seems to have
9 >> changed.
10 >
11 > I wonder if it could be related to parallel initialization of disks.
12
13 Apparently so.
14
15 > I think there's a kernel toggle for that.
16
17 I'll look into it.
18
19 >> Are we really expected now to set up an initrd just so that the
20 >> kernel can find the root partition??
21 >
22 > As far as I know, the answer is "yes".
23 >
24 > FWIW, I always resisted making an initrd until very recently, but
25 > wanted to use UUID in my bootup on my new system. I used this command
26 > (which I re-run whenever I deploy a new kernel):
27 >
28 > dracut -H -o i18n -o resume -o usrmount --force /boot/initramfs.img
29 >
30 > And then added one line to my grub2 config:
31 >
32 > initrd /initramfs.img
33 >
34 > and it just simply works... though it's still a bit of black magic to
35 > me,
36
37 It's not really that complex. There are two ways to get the kernel to
38 use an initial ram fs as root: the bootloader loads the initial ram
39 image into ram or it gets linked into the kernel when the kernel is
40 built.
41
42 > and every time I reboot I feel a bit of nervousness when I see
43 > "Loading initial ramdisk..." and don't breathe until it succeeds. :)
44
45 I've got nothing against initramfs -- on my embedded systems root
46 starts out as an initramfs and stays that way. I just don't see why I
47 should be forced to have one.
48
49 --
50 Grant