Problem with network shared hdd

Fill · 01-03-2010, 09:54 AM

Hi,

I have a 1TB Samsung hard disk placed inside of one of those boxes that transform internal SATA hard disks into an external USB disk. The disk is plugged via USB cable to my home server which runs Debian 5.0.3 and linux kernel 2.6.26-2-686.

This disk is used to save some backups and files that we need to access often on different computers of the local network (using WinSCP). Today, something weird is happening. While updating some backups from one Windows Vista box with WinSCP, the connection to the server is lost and all the directories on the disk disappear. The disk kind of freezes.

so .. checking on system log, I find this:

Quote:

Jan 3 15:20:06 BOX4 kernel: [675850.568009] __journal_remove_journal_head: freeing b_frozen_data
Jan 3 15:20:06 BOX4 kernel: [675850.568009] __journal_remove_journal_head: freeing b_committed_data
Jan 3 15:20:06 BOX4 kernel: [675850.568009] __journal_remove_journal_head: freeing b_frozen_data
Jan 3 15:20:06 BOX4 kernel: [675850.568009] __journal_remove_journal_head: freeing b_frozen_data
Jan 3 15:20:06 BOX4 kernel: [675850.568009] __journal_remove_journal_head: freeing b_frozen_data
Jan 3 15:20:06 BOX4 kernel: [675850.568009] __journal_remove_journal_head: freeing b_frozen_data
Jan 3 15:20:06 BOX4 kernel: [675850.568009] Buffer I/O error on device sda1, logical block 122061314
Jan 3 15:20:06 BOX4 kernel: [675850.568009] lost page write due to I/O error on sda1

The disk is not one of those old hard disks, I bought it less than a year ago. I tried to run some diagnostic tools like SMART but self test features and logging is not supported (maybe because it's USB). The disk inside the external box is a SAMSUNG HD103SI. I don't know what to do and we really need this up and running. Also, obviously I mounted it again and tried to perform the SSH backup once again, but it failed again.

Another fact that I would like to refer is that I have a second 1TB SAMSUNG HD154UI hard disk which is only used to store movies and music, and is up and running without any problems, so I wouldn't say it's controllers problem.. !

What should I do now?

Thanks in advance,
Fill

AsusDave · 01-03-2010, 10:06 AM

With which file system did you have the drive formatted? Did you lose power around the time this happened?

HTH
Dave

Fill · 01-04-2010, 10:30 AM

Hi AsusDave, thanks for the fast reply.

In fact we've had some power problems arround our hood, the situation repeated for 2 or 3 consecutive days. I remember one of them was in 24th december at night :X

Did the disk got corrupt because of the power failures? It is plugged in a surge-protector, I think that would be enough to have my disk out of troubles (no surges = no problems)... hummm, maybe the sudden shutdown caused by the power loss made something wrong ..?

btw, just to let you know, my disk is using ext3 filesystem

AsusDave · 01-05-2010, 10:10 AM

It "could" have become corrupted. Ext3 is a journaled file system meaning it tends to handle power outages pretty well though. I would try running fsck to see if it can fix any errors in the file system.

HTH
Dave

Fill · 01-05-2010, 03:49 PM

Hey,

How do I use fsck? Is there any tutorial or something like that to help me?

AsusDave · 01-06-2010, 07:10 AM

Take a look at the man page, it has a bunch of really cool options.

The short version is basically

Code:

fsck <insert options here> <insert device to scan here> <insert additional options here>

I think "don't quote me" there is also a -F option you can run at shutdown to force a fsck of your disks on reboot. Depending on your disk sizes, that might mean a while without your server.

HTH
Dave

Fill · 01-06-2010, 01:22 PM

Hey,

As you said, I've run fsck, and I've came up with this:

Quote:

BOX4:~# fsck -a -t ext3 /dev/sda1
fsck 1.41.3 (12-Oct-2008)
networkdisk: recovering journal
networkdisk has been mounted 49 times without being checked, check forced.
networkdisk: 125279/61054976 files (1.6% non-contiguous), 62563739/244190000 blocks
BOX4:~#

What does that last line exactly mean? do you think I can remount the disk and put it back online?

See you arround,
Fill

michaelk · 01-07-2010, 04:49 AM

Quote:

networkdisk: 125279/61054976 files (1.6% non-contiguous), 62563739/244190000 blocks

This line tells you how much the disk is fragmented and not a problem. The journal appears to be recovered and looks like it should be ready to be mounted and put back online.

Brown outs and dropouts can be just as bad or worse then power surges which surge protectors can not provide any protection from. If you experience lots of power problems an UPS (uninterruptible Power Supply) is recommended.

AsusDave · 01-07-2010, 06:20 AM

I concur with michael, it looks good to go.

I also agree UPS is a really good idea.
Just be glad you were using a journaled file system or you would have likely lost some or all of your data.

HTH
Dave

AsusDave · 01-07-2010, 06:20 AM

I concur with michael, it looks good to go.

I also agree UPS is a really good idea.
Just be glad you were using a journaled file system or you would have likely lost some or all of your data.

HTH
Dave

Fill · 01-07-2010, 02:12 PM

Hi,

Already got an UPS, actually I misread your information and what I wanted to say is that the disk was always plugged on an UPS (we here have almost our pc's on a UPS based system). I mounted it, but unfortunately the situation remains the same. Any more suggestions?

If there isn't anything else I can do, I'll move all the data to another temporary disk and buy a new one asap (Assuming that this one just died).

Once again, thanks
Fill

michaelk · 01-07-2010, 02:43 PM

Do you still see the same errors?
Can you mount the drive?

Fill · 01-07-2010, 04:28 PM

Yeah, I was able to mount it. When I started copying files again, I got this:

Quote:

Jan 7 22:09:01 BOX4 kernel: [1045985.561449] sd 4:0:0:0: [sda] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE,SUGGEST_OK
Jan 7 22:09:01 BOX4 kernel: [1045985.561449] sd 4:0:0:0: [sda] Sense Key : Medium Error [current]
Jan 7 22:09:01 BOX4 kernel: [1045985.561449] sd 4:0:0:0: [sda] Add. Sense: Unrecovered read error
Jan 7 22:09:01 BOX4 kernel: [1045985.561449] end_request: I/O error, dev sda, sector 77054935
Jan 7 22:09:01 BOX4 kernel: [1045985.713392] sd 4:0:0:0: [sda] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE,SUGGEST_OK
Jan 7 22:09:01 BOX4 kernel: [1045985.713392] sd 4:0:0:0: [sda] Sense Key : Medium Error [current]
Jan 7 22:09:01 BOX4 kernel: [1045985.713392] sd 4:0:0:0: [sda] Add. Sense: Unrecovered read error
Jan 7 22:09:01 BOX4 kernel: [1045985.713392] end_request: I/O error, dev sda, sector 77054935
Jan 7 22:09:01 BOX4 kernel: [1045985.868786] sd 4:0:0:0: [sda] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE,SUGGEST_OK
Jan 7 22:09:01 BOX4 kernel: [1045985.868794] sd 4:0:0:0: [sda] Sense Key : Medium Error [current]
Jan 7 22:09:01 BOX4 kernel: [1045985.868799] sd 4:0:0:0: [sda] Add. Sense: Unrecovered read error
Jan 7 22:09:01 BOX4 kernel: [1045985.868805] end_request: I/O error, dev sda, sector 77054935

michaelk · 01-08-2010, 06:55 PM

Its either the hard drive failing or the chassis has a hardware problem.

Fill · 01-09-2010, 09:55 AM

which means that there is nothing wrong with my system, right? What would you recommend me? Buy a new disk? Isn't there anything else I can do to try to get it back to work?

[Edited]
The SSH copy is now perfectly working with the temporary hard disk that I'm using now...