[linux-lvm] Re: One PV destroyed - now the whole VG isn't working anymore

[linux-lvm] Re: One PV destroyed - now the whole VG isn't working anymore

[Samuel Leuthold]

Hello

I have the same Problem as Sebastian. I have 3 disks in my LV. The third one
has badblocks. The Data isn't nessecary on this disk.
So can i release the third disk from the LV, reformat and rebind into it
without loosing the data on disk1 and 2?? And what about the reiser? Is it
consistent after this "hack"??

disk1 /dev/hdb
disk2 /dev/hdc
disk3 /dev/hdd  <-- the faulty one and the last joined to the LV

Can i do it with these commands?? What is different? Whats the meaning of
the "0" in storage\0 ??

mkdir /etc/lvmtab.d
cp /etc/lvmconf/storage.conf /etc/lvmtab.d/storage
echo -ne "storage\0" > /etc/lvmtab
vgchange -ay storage

Please help... i don't want to lose 240GB!

Thanks for help!
Samuel

<<SNIP>>--------------------------------------------------------------------
-------------
Message: 5
Date: Mon, 3 Nov 2003 11:05:55 +0100
From: "Heinz J . Mauelshagen" <mauelshagen@sistina.com>
To: linux-lvm@sistina.com
Subject: Re: [linux-lvm] One PV destroyed - now the whole VG isn't working
anymore
Reply-To: linux-lvm@sistina.com


Sebastian,

you either want to use LVM2/device-mapper and try the "vgchange -P ..."
option to activate partially available VGs

-or-

a quick hack to get LVM1 going:

mkdir /etc/lvmtab.d
cp /etc/lvmconf/storage.conf /etc/lvmtab.d/storage
echo -ne "storage\0" > /etc/lvmtab
vgchange -ay storage

If you have those contents in /etc/lvmtab and /etc/lvmtab.d/storage restored
anyway, "vgchange -ay storage" is all you need to do.

Either way presumes, that enough logical extends are accessable fine. Backup
your data and recreate the VG afterwards.

Regards,
Heinz    -- The LVM Guy --
<<SNAP>>--------------------------------------------------------------------
-------------

Re: [linux-lvm] Re: One PV destroyed - now the whole VG isn't working anymore

[Rickard Olsson]

Samuel Leuthold wrote:

> So can i release the third disk from the LV, reformat and rebind into it
> without loosing the data on disk1 and 2?? 

Yes. But, if you are using striping instead of simple concatenation, you 
are, as the saying goes, toast.

However, it may be possible to salvage most of the data from the third 
disk too.

Method #1.
Use   dd conv=noerror   to raw copy data off the bad disk to a fourth, 
good disk. Swap the bad disk for the new one.

Method #2.
If there's free space on the other two disks (or you can add a fourth 
disk), you can try   pvmove --ignore_read_errors   to move as many 
extents as possible off the bad disk onto the good ones.

> And what about the reiser? Is it
> consistent after this "hack"??

No, in all cases you will need to run reiserfsck, most likely with the 
-rebuild-tree option. But seeing as your disk has badblocks, your FS is 
already in a bad state.

 > Can i do it with these commands??

If your metadata is intact, you won't need most of that example. You can 
search the list archive for my conversation with Heinz a little over a 
month or so ago, we went through this exact thing. :-)

> Whats the meaning of
> the "0" in storage\0 ??

Think of it as an end-of-line character (it's a null-terminated string).

    / Rickard Olsson,IT-Konsult/
   / Telefon: +46 70 635 01 42/
  / http://www.webhackande.se/

Re[2]: [linux-lvm] Re: One PV destroyed - now the whole VG isn't working anymore

[Spam]

> Samuel Leuthold wrote:

>> So can i release the third disk from the LV, reformat and rebind into it
>> without loosing the data on disk1 and 2?? 

> Yes. But, if you are using striping instead of simple concatenation, you
> are, as the saying goes, toast.

> However, it may be possible to salvage most of the data from the third
> disk too.

> Method #1.
> Use   dd conv=noerror   to raw copy data off the bad disk to a fourth,
> good disk. Swap the bad disk for the new one.

  I  would  use  dd_rescue  instead  of dd. dd_rescure is specifically
  made  for copying data off broken disks. If you have bad sectors you
  will  be  much  better  off with dd_rescue. IF the disk stalls while
  copying  you  can  tell dd_rescue to copy backwards to avoid reading
  from that sector again.

  If  you  do  not have to much bad sectors you should be able to save
  most data from your filesystem.

> Method #2.
> If there's free space on the other two disks (or you can add a fourth
> disk), you can try   pvmove --ignore_read_errors   to move as many 
> extents as possible off the bad disk onto the good ones.

>> And what about the reiser? Is it
>> consistent after this "hack"??

> No, in all cases you will need to run reiserfsck, most likely with the
> -rebuild-tree option. But seeing as your disk has badblocks, your FS is
> already in a bad state.

  Be  sure to use most current reiserfsck. Latest is 3.6.11 and can be
  downloaded from: ftp://ftp.namesys.com/pub/reiserfsprogs/

 >> Can i do it with these commands??

> If your metadata is intact, you won't need most of that example. You can
> search the list archive for my conversation with Heinz a little over a
> month or so ago, we went through this exact thing. :-)

>> Whats the meaning of
>> the "0" in storage\0 ??

> Think of it as an end-of-line character (it's a null-terminated string).