[mythtv-users] Recovering from a nasty crash

[Dale Pontius]

On 08/27/10 03:25, Russell Gower wrote:
>>
>> If anyone else has been in this situation and has any advice, I'd
>> appreciate hearing it.
> 
> I've been in a similar situation, it wasn't a Myth system rather my main file server, This is what I would do to start with.
> 
> Since I presume you have plenty of spare space on that 1TB drive I would create two partitions both slightly larger than the number of blocks required for the 160G and 200g drives (if your not confutable calculating the number of blocks required I would go with 170G and 210G) Then one at a time using the channel you suspect is ok, do a block level copy from the partition on each of the old disks to the partition on the new one using dd, then once both have been copied you should be able to rebuild the LVM and mount the filesystem.
> 
> If you would like more specific advice please free to contact me off list and i'll try and work out the exact commands you need to type.
> 
I'd already done pretty much what you suggested.  My first attempt
failed pretty miserably, only getting some critical information off of
the machine.

Yesterday evening I replugged one IDE connector on the motherboard,
(couldn't reach the other without some major dismantling) and swapped
the drive cables, so each drive was on the opposite channel.  I was kind
of wondering if this wasn't a motherboard/chipset problem instead of a
drive problem.  I then began going after information more carefully.  I
managed to grab all of /home, including the fact that all of my nightly
MythDB backups go into a subdirectory of /home/mythtv.  I grabbed all of
/usr/local, I already had /etc, and a few other select things.  Then I
went after /media, and actually got all of it.  At the same time, in
another window I had a root-tail on /var/log/messages, and saw no
hardware errors whatsoever.  After this I managed to get /var/lib/mysql
- the mysql binary runtime database.  I may as well give this a try
before reconstructing from scratch and a backup.

Things had gone so well that I decided to try another simple root copy
off of the previously-bad 160G drive.  The hardware errors started
almost immediately, except of course they'd moved from ata5 to ata6.  So
clearly the 160G drive is toast.

With a backup of /media, I managed to back the 200G drive out of LVM2
and am in the process of installing Gentoo on it.  (Currently booted,
running "emerge -atuvDN system" compiling (15 of 44) gcc.  I hope to
have the machine back in service by Monday.

This is an old socket 939 board, and this spring at Flea at MIT I
managed to pick up a new in-the-box Athlon64 CPU for this board, a bit
faster than my current one.  (3500 vs 3000)  Normally I wouldn't bother
with such a minor speed increase, but for $5, new in the box...  I hope
to install the new CPU before putting it back in service.

Thanks for your advice,
Dale