1 |
Francesco Talamona wrote: |
2 |
> |
3 |
> First of all, thanks for sharing. |
4 |
> |
5 |
> I used to think xfs was overkill for /boot, but the procedure described |
6 |
> is quite straightforward. |
7 |
> |
8 |
> There are two things I don't understand: |
9 |
> |
10 |
> 1) why do you delete xfsdump and xfsrestore in /xfsrestore/usr/bin/ just |
11 |
> extracted to link them to /xfsrestore/sbin |
12 |
> |
13 |
> 2) the use of df.out isn't clear to me, isn't the dump file name enough |
14 |
> to know what is in there? |
15 |
> |
16 |
|
17 |
1) The symlinks are broken if the package is extracted anywhere other |
18 |
than /. I recreated 'em to point where they should (I recall they were |
19 |
needed, as some of the ancillary programs break if they are missing or |
20 |
broken). |
21 |
|
22 |
2) The df.out is so you know that (say) usr.0.dmp should be restored to |
23 |
a device called (say) /dev/sda6. This will avoid the need to edit |
24 |
restored /etc/fstab (or the need to boot into single user mode and fix |
25 |
it). the other point is if you are reusing the same disk setup (assuming |
26 |
a software issue is requiring the restore), then checking df.out ensures |
27 |
that you recover the system using the same partitions for the |
28 |
filesystems as you had pre-restore. |
29 |
|
30 |
Cheers |
31 |
|
32 |
Mark |
33 |
|
34 |
P.s: You are quite correct that xfs is overkill for /boot. However I |
35 |
just found it easier to xfs everything (otherwise I'd have to use |
36 |
different dump programs depending on what I was backing up etc... ). To |
37 |
me this is more important than the fact that it wastes disk space a bit |
38 |
(my /boot uses a 128M partition but only gets 93M to actually use...and |
39 |
it uses 11M of that! - but disks are quite big now...) |
40 |
-- |
41 |
gentoo-user@g.o mailing list |