1 |
In linux.gentoo.user, Tanstaafl wrote: |
2 |
> On 2013-09-29 4:09 PM, Alan McKinnon <alan.mckinnon@×××××.com> wrote: |
3 |
>> On 29/09/2013 19:59, Tanstaafl wrote: |
4 |
>>> I've been told that this shouldn't be a big deal... while I am a |
5 |
>>> (barely) passable linux sys admin |
6 |
>> |
7 |
>> Allow me to forward an opinion. The above is not true, not even close. |
8 |
>> |
9 |
>> Don't knock yourself, you don't deserve it :-) |
10 |
> |
11 |
> Lol!!! At first I thought you were saying that it wasn't true that |
12 |
> merging /usr into / shouldn't be a big deal - and I was about to start |
13 |
> gnashing my teeth (again). |
14 |
> |
15 |
> Thanks Alan, your words are very kind... and I'll just leave it at |
16 |
> that... ;) |
17 |
|
18 |
I've just changed one of my machines so that /usr is now part of the |
19 |
root filesystem. Like you, I had a separate /usr filesystem. Unlike you |
20 |
I've been running an initramfs for many years because: |
21 |
|
22 |
a) I'm running laptops and like them to have pretty graphical boot |
23 |
screens and no "ugly writing" appearing during the boot sequence. It's |
24 |
silly, I know, but it still looks pretty. The initramfs will start up |
25 |
"bootsplash" 8-) |
26 |
|
27 |
b) The important reason I need an initramfs is that I have my root |
28 |
filesystems on LVM partitions (except for my ARM servers). I've never |
29 |
has a scrap of trouble with the genkernel initramfs builds, despite |
30 |
myriad updates over the years. I've had minor niggles with display but |
31 |
nothing critical. |
32 |
|
33 |
So while I've run an initramfs for many years, now it has had to mount |
34 |
/usr before the "pivot_root" command. This has led to the problem that |
35 |
/usr is no longer able to be fscked because it is already mounted, and I |
36 |
cannot for the life of me, get the genkernel initramfs to fsck the /usr |
37 |
filesystem before mounting. I've had to manually fsck the /usr |
38 |
filesystem by running my minimal install CD. There are probably ways to |
39 |
do this (like fscking /usr on shutdown, which I couldn't get working) |
40 |
but I'm sick of looking for them. I've bit the bullet and changed things |
41 |
over. It went without a hitch. |
42 |
|
43 |
Here's what I did: |
44 |
|
45 |
I added a new LVM volume group and added a "slash" filesystem (10Gb), a |
46 |
"usrsrc" filesystem for my kernels (10Gb), a "portage" filesystem (3Gb), |
47 |
a "distfiles" filesystem (15Gb) and a "packages" filesystem (10Gb). |
48 |
Because these are on LVM they can be adjusted upwards or downwards |
49 |
depending on usage. I updated /etc/default/grub so that the new kernel |
50 |
command line will find my new "slash" LVM volume, and ran the grub2 |
51 |
installer to make the change valid. |
52 |
|
53 |
I then shut down the machine, booted my minimal install CD, used LVM to |
54 |
find my filesystems. I then mounted my new "slash" and mounted the new |
55 |
filesystems. |
56 |
|
57 |
I also decided to move portage, distfiles and packages to the old /var |
58 |
partition but to do so I first had to mount them in their old positions |
59 |
on /usr/portage /usr/portage/distfiles etc... Once done, I mounted the |
60 |
old "slash" and the old "/usr" (with included distfiles and packages and |
61 |
portage) then did the "cp -av <old hierarchy> <new hierarchy>". It was |
62 |
then possible to unmount distfiles, packages and portage and then move |
63 |
them to /var (mount /var and mkdir /var/portage /var/distfiles and |
64 |
/var/packages) I altered the new "slash" fstab. I then rebooted without |
65 |
a hitch. Oh, I also had to update /etc/portage/make.conf and the |
66 |
"make.profile" symlink to reflect the change. |
67 |
|
68 |
It seems complicated but every step was logical. Having my root |
69 |
filesystem on LVM has made the change more complicated than it should |
70 |
have been, but it still was quite easy to do and downtime was minimal. |
71 |
|
72 |
I don't feel like I've been "forced" to do anything. I'm grateful for |
73 |
the Gentoo devs and their hard work over the years. This upstream change |
74 |
is just a small bump in the long Gentoo road. If I didn't agree with the |
75 |
change then it would be up to me to find a way to get my system to work |
76 |
without an initramfs, not the Gentoo Devs... after all, this IS open |
77 |
source. Be grateful that the Gentoo Devs are still willing to volunteer |
78 |
their time building this great distribution. |
79 |
|
80 |
-- |
81 |
Regards, |
82 |
Gregory. |