usb hard drive diagnosis

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

usb hard drive diagnosis

Johannes Wiedersich
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I'm a bit concerned about the health status of my usb based hard disk
backup. One of my recent backups (rsync) prematurely exited with I/O
errors in syslog. I fsck'ed the drive, fixing some 2000 errors like

/-------------
Free inodes count wrong for group #1535 (11199, counted=11196).
Fix? yes

Directories count wrong for group #1535 (416, counted=419).
Fix? yes
\-------------

and bitmap differences.

I did a new backup to the drive and it seems everything went fine. I
also checked some of the data on the disk with git-fsck for consistency.

What concerns me a bit is that after each backup, when I force a new
fsck, I get this:

/-------------
e13-v15:~# e2fsck -f /dev/disk/by-label/maxtor
e2fsck 1.41.3 (12-Oct-2008)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 3A: Optimizing directories
Pass 4: Checking reference counts
Pass 5: Checking group summary information

maxtor: ***** FILE SYSTEM WAS MODIFIED *****
maxtor: 5059092/61063168 files (0.7% non-contiguous), 97098262/122096000
blocks
\-------------

(the filesystem was modified line). I don't recall seeing this line too
often in the past.
Does this indicate problems with the disk?

Unfortunately, smartmontools won't work with usb and I hesitate to break
the warranty seal to mount the disk directly.

Are there any other tools to check the health status of a usb disk?

Is fsck.ext3's badblocks option (-c) useful for usb disks or is it
somehow inefficiently interfering with the usb transmission and/or the
disk's internal controller?

Thanks for helping to make me sleep better ;-)

Cheers,

Johannes

NB: The disk contains several hundred GB of data and each backup is
created with hard linking to the previous one.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkmBcbsACgkQC1NzPRl9qEV1QgCeLS2ijVvi0yAYr5paArhLg7iZ
EPsAn2F8IfhIBqw/20bI6akmiMlf+kT+
=AZOQ
-----END PGP SIGNATURE-----


--
To UNSUBSCRIBE, email to [hidden email]
with a subject of "unsubscribe". Trouble? Contact [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: usb hard drive diagnosis

David E. Fox
On Thu, Jan 29, 2009 at 1:07 AM, Johannes Wiedersich
<[hidden email]> wrote:

> I'm a bit concerned about the health status of my usb based hard disk
> backup. One of my recent backups (rsync) prematurely exited with I/O
> errors in syslog. I fsck'ed the drive, fixing some 2000 errors like

Ow. That seems like a lot. :(

> maxtor: ***** FILE SYSTEM WAS MODIFIED *****
> maxtor: 5059092/61063168 files (0.7% non-contiguous), 97098262/122096000
> blocks

Tnat in and of itself doesn't indicate that there is a problem with
the drive. I've frequently (well maybe hopefully not that often) had
to run that kind of filesystem check - even in 'manual fsck' mode
where the system only will partially boot up and not continue until
fhe filesystem is correctly checked through all five passes.

I've had a few drives in service for years (retired ibm deskstar 30gb
etc.) they should still work even though I've had to fsck them a few
times.


> Unfortunately, smartmontools won't work with usb and I hesitate to break
> the warranty seal to mount the disk directly.

Hmm perhaps this is true, since I don't have any USB hard drive media,
but I would think they fall into two distinct types:

* flash (RAM) drives
* enclosures that are basically a standard sata or ide drive with a
power and usb cable added

 For #2 I'm not sure why you couldn't treat it more like a regular HD
except for the  USB part gets in the way (it's slower in terms of file
I/O through a USB bus instead of a more standard ide or SATA
connection.

But why wouldn't things like smartmontools work through the USB?


--
To UNSUBSCRIBE, email to [hidden email]
with a subject of "unsubscribe". Trouble? Contact [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: usb hard drive diagnosis

Johannes Wiedersich-3
David Fox wrote:
> But why wouldn't things like smartmontools work through the USB?

From [1]:
"As for USB and FireWire (IEEE 1394) disks and tape drives, the news is
not good. "

It seems to be improving, though.

Cheers,

Johannes

[1] http://smartmontools.sourceforge.net/faq.html#testinghelp




signature.asc (268 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: usb hard drive diagnosis

Hugo Vanwoerkom
In reply to this post by David E. Fox
David Fox wrote:
<snip>

> Hmm perhaps this is true, since I don't have any USB hard drive media,
> but I would think they fall into two distinct types:
>
> * flash (RAM) drives
> * enclosures that are basically a standard sata or ide drive with a
> power and usb cable added
>
>  For #2 I'm not sure why you couldn't treat it more like a regular HD
> except for the  USB part gets in the way (it's slower in terms of file
> I/O through a USB bus instead of a more standard ide or SATA
> connection.
>
> But why wouldn't things like smartmontools work through the USB?
>
>

I've posted this before, but I have 2 USB HDD enclosures. I bought the
enclosures and the HDD's separately and put them together myself. 1 with
a SATA disk and 1 with an IDE disk. The latter I can get results with:

smartctl -t long -d sat /dev/sda -T permissive
smartctl -d sat /dev/sda --all -T permissive

The former gives me an error when I do that. Funny thing is that I can
only get good results with Debian's (Sid) kernel:

hugo@debian:/$ uname -a
Linux debian 2.6.26-1-686 #1 SMP Sat Jan 10 18:29:31 UTC 2009 i686 GNU/Linux

*not* with my own kernel of the same version. Never found an explanation
of that either.

Hugo


--
To UNSUBSCRIBE, email to [hidden email]
with a subject of "unsubscribe". Trouble? Contact [hidden email]