Re: HELP! Re: How to fix I/O errors?

[David Christensen <dpchrist@holgerdanske.com>]

On 02/07/17 23:37, Marc Shapiro wrote:
> How it went is not well.

> David Christensen wrote:
>> Run memtest86+ for 24+ hours to verify that you don't have a memory
>> problem.

Did you test the memory?  If not, test it now just to be sure.


>> Use SeaTools to wipe the new 1 TB drive and run the short and long
>> tests.  Stop if anything fails.
> I tested the new drive with SeagateTools and it
> was fine.

Please confirm that you wiped the 1 TB recovery drive.


> Then I made a clonezilla live CD and booted from it.  It stopped
> on the first read error with a message saying to restart using the rescue
> option.  I did that.  After 5 hours it finished without mentioning any
> errors.
>
> I tried to boot to the old disk (since it was still wired that way).  I got
> dropped int a maintenance shell with fs errors in /dev/sda4 which is the
> physical volume for all my LVM logical volumes -- /usr, /var, /home and
> /temp.  It says to run fsck manually.
>
> I decided to try the new drive, so I changed the cables and re-booted.
>
> Maintenance shell, again.
>
> / mounted clean
>
> lvm started
>
> /home fs has errors run fsck (at this point, I'm afraid to try it)
>
> /var, /usr, and /tmp all say that the superblock can not be read, or is
> invalid.  Try running
>
> e2fsck -b 8193 <device>
> or
> e2fsck -b 32768 <device>
>
> Which do I use?
>
> How did trying to clone the disk nake such a mess of BOTH disks?

Don't blame Clonezilla.  Everything is decaying -- you, me, those hard 
drives, etc..  With that in mind, do the most precious operations first 
-- because in 1 second, 1 minute, 1 hour, 1 day, 1 month, 1 year, 1 
decade, 1 century, whatever, the data will be inaccessible without 
extraordinary means.


Forget about booting off the failing 1 TB disk.  Disconnect it for now.


Forget about booting off the 1 TB recovery disk.  It should now contain 
whatever blocks Clonezilla was able to recover.  It is now in a state 
analogous to Swiss cheese.  Disconnect it for now.


> Any help getting a working system again will be greatly appreciated.

On the computer you use for e-mail, start an administration log folder 
for the machine in question.  Start a log.txt file and take notes.  Cut 
and paste what you can.  Photograph screens and transcribe what you 
can't.  Collect important files.  Put it all into a version control system.


>> I'd do a fresh install on a 16+ GB SSD (USB flash drives also
>> work).

Install SSH when you build the new system drive.


Use ssh(1) to log in from your e-mail computer.  Consider using 
script(1) to capture your console sessions, and scp(1) to copy out the 
files.  Read fsck(8) and consider your moves carefully.  Reconnect the 1 
TB recovery disk and see what fsck can recover.


David