1 |
On Wednesday 24 January 2007 20:37, jcd wrote: |
2 |
|
3 |
[snip] |
4 |
|
5 |
> "Everything was fine" mean; I created partition and then formatted it |
6 |
> without any errors or warnings. There are messages from syslog: |
7 |
> --------------------------------------------------------------------- |
8 |
>-- Jan 22 23:43:16 localhost EXT3 FS on sdb1, internal journal |
9 |
> Jan 22 23:43:16 localhost EXT3-fs: mounted filesystem with ordered |
10 |
> data mode. |
11 |
> --------------------------------------------------------------------- |
12 |
>-- Then I copied my data to this new partition. I could access this |
13 |
> data from new partition without any problems. Next day: |
14 |
|
15 |
OK, so we will assume that the data was written correctly |
16 |
|
17 |
> --------------------------------------------------------------------- |
18 |
>-- Jan 23 10:23:46 localhost VFS: Can't find ext3 filesystem on dev |
19 |
> sdb1. |
20 |
> --------------------------------------------------------------------- |
21 |
>-- |
22 |
> |
23 |
> > It looks like when you moved the data onto the new partition, it |
24 |
> > got written somewhere on the disk. However, the kernel's idea of |
25 |
> > how the partitions are laid out at that time and what fdisk just |
26 |
> > wrote to the disk probably don't agree and the kernel had got it |
27 |
> > wrong.... This does happen when you delete two or more partitions |
28 |
> > and create one large one. |
29 |
> |
30 |
> Why it can happen when replacing two partitions with large one? |
31 |
|
32 |
First thing to know, is that the PC has the most insane internal design |
33 |
of any electronic device ever made anywhere in the world at any time, |
34 |
ever. (Well, Thomson aircraft radios are actually worse, but you get |
35 |
the idea...). The result is that not everything makes sense... When the |
36 |
kernel boots, it reads the partition table off disk and knows that the |
37 |
first partition starts at cylinder 0 and the second partition starts at |
38 |
say cylinder 2000. The kernel doesn't update this information when you |
39 |
run fdisk, so if you delete two partitions and create one big one, the |
40 |
kernel can get confused. It's not hard to fix on the PC, but Linux runs |
41 |
on 20 architectures that are not all as crazy as Intel PCs, which might |
42 |
be why this oddity is still there are 15 years. Redhat have a utility |
43 |
called partprobe that gets everything back in sync after using fdisk, |
44 |
but I have yet to find it in Portage |
45 |
|
46 |
> I tried gpart with this output: |
47 |
> --------------------------------------------------------------------- |
48 |
>-- #gpart /dev/sdb |
49 |
> Begin scan... |
50 |
> Possible partition(Windows NT/W2K FS), size(40959mb), offset(0mb) |
51 |
> Possible partition(Linux ext2), size(197512mb), offset(40959mb) |
52 |
> End scan. |
53 |
|
54 |
[snip] |
55 |
|
56 |
According to this you have an ext2/3 partition as the SECOND partition, |
57 |
not the first, and it does not cover the whole disk. |
58 |
|
59 |
Are you absolutely sure you pressed "w" in fdisk after creating the |
60 |
partitions? It sure looks to me like your changes were not written to |
61 |
disk. Try mounting /dev/sdb2 |
62 |
|
63 |
alan |
64 |
-- |
65 |
gentoo-user@g.o mailing list |