[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: where did my ata drives go?

On Fri, Sep 10, 2010 at 09:54:38AM -0400, Stephen Powell wrote:
> On Thu, 09 Sep 2010 22:58:30 -0400 (EDT), Rick Pasotto wrote:
> > 
> > Last weekend I did a bunch of updates (to testing) and rebooted after
> > almost six months.
> You really should upgrade more often than that when running testing.
> > Somewhere in that process the ata drive got UUIDs assigned to the
> > partitions and /etc/fstab was modified.
> I suspect what happened is a migration from kernel 2.6.32-3-xxx
> to kernel 2.6.32-5-xxx, which installs linux-base, which tries to
> convert system files such as /etc/fstab,
> /etc/initramfs-tools/conf.d/resume, and other system files
> to use UUIDs whenever possible, rather than /dev/hdax references.
> The UUIDs were already assigned to the partitions.  The partitions
> just weren't being mounted by UUID, they were being mounted by
> device name.  Now they are mounted by UUID.
> > Now they won't mount.
> Yes, they are still being mounted.

No, they are not! They do not show up with 'df'.

> They just have different device names now.  That is due to a change in
> device drivers between the 2.6.32-3-xxx kernel and the 2.6.32-5-xxx
> kernel.  The 2.6.32-3-xxx kernel uses the traditional driver for
> traditional IDE hard disks, also known as ATA (AT attachment) or PATA
> (parallel AT attachment). This driver uses device names of the form
> /dev/hda, /dev/hdb, etc. The 2.6.32-5-xxx kernel uses a newer driver
> for these disks which uses SCSI emulation.  The newer driver uses SCSI
> (Small Computer System Interface) device names even for PATA disk
> drives.  Thus, the device names are called /dev/sda, /dev/sdb, etc.
> It's a different naming convention for the same drives.

Yes, I now remember that message coming through when I installed the
latest kernel.

> > I used to mount /dev/hdb1 and /dev/hdb5 but those don't exist
> > anymore.
> Right.  As I explained above.  But it you were to boot your old
> kernel, the 2.6.32-3 kernel, you would see the old device names
> re-appear again.  The reason that linux-base tries to migrate system
> files such as /etc/fstab to UUID-based mounting is so that the mount
> will succeed regardless of which kernel you boot.

Haven't tried that yet, but I will.

> > Using tune2fs I can access them as /dev/sdc1 and /dev/sdc5. The
> > UUIDs match what got written to /etc/fstab. I was able to assign
> > labels to them using tune2fs but they still refuse to mount.
> Not by their old names, no.  The devices have different names under
> the new kernel.
> > How can I access those partitions?
> You already are accessing them, but with different names now. You
> haven't lost any data.

The original line in /etc/fstab was:

/dev/hdb1            /hd0            ext3    defaults        0       0

That line got commented out and this line was added:

UUID=03c23684-dea8-458d-b04b-0ae8a056cb0d            /hd0            ext3    defaults        0       0

Using tune2fs I added the label 'hdb1' and added this line to /etc/fstab:

LABEL=hdb1            /hd0            ext3    defaults        0       0

'mount /hd0' DOES NOT WORK! It gives this error message:

mount: special device LABEL=hdb1 does not exist

'tune2fs -l /dev/sdc1' gives:

tune2fs 1.41.12 (17-May-2010)
Filesystem volume name:   hdb1
Last mounted on:          <not available>
Filesystem UUID:          03c23684-dea8-458d-b04b-0ae8a056cb0d
Filesystem magic number:  0xEF53
Filesystem revision #:    1 (dynamic)
Filesystem features:      has_journal filetype sparse_super large_file
Filesystem flags:         signed_directory_hash 
Default mount options:    (none)
Filesystem state:         clean
Errors behavior:          Continue
Filesystem OS type:       Linux
Inode count:              2443200
Block count:              4883752
Reserved block count:     244187
Free blocks:              627830
Free inodes:              2399380
First block:              0
Block size:               4096
Fragment size:            4096
Blocks per group:         32768
Fragments per group:      32768
Inodes per group:         16288
Inode blocks per group:   509
Last mount time:          Mon Jul 19 11:40:39 2010
Last write time:          Thu Sep  9 21:38:18 2010
Mount count:              90
Maximum mount count:      38
Last checked:             Thu Jun 30 03:47:39 2005
Check interval:           15552000 (6 months)
Next check after:         Tue Dec 27 02:47:39 2005
Reserved blocks uid:      0 (user root)
Reserved blocks gid:      0 (group root)
First inode:              11
Inode size:	          128
Journal inode:            8
Journal backup:           inode blocks

The label (volume name) is there. The UUID matches. I get the same
(non-)results whether I try mounting with the label or the UUID.

"Victory goes to the player who makes the next-to-last mistake,"
said chess master Savielly Grigorievitch Tartakower.
    Rick Pasotto    rick@niof.net    http://www.niof.net

Reply to: