From: J.O. Aho on
Ohmster wrote:
> noi <noi(a)siam.com> wrote in news:EeXTg.6607$TV3.5327
> @newssvr21.news.prodigy.com:
>
>> Second do not panic.
>> With FC3 as the primary system, boot from the FC5 DVD
>>
>> boot: linux rescue
>> # you get the prompt
>>
>> $ chroot /mnt/sysimage
>>
>> $ ls ~
>>
>> # you should still see your data
>>
>> If you can see your data then you can disconnect the FC3
>> and reinstall FC5 without the LVM.
>
> No can do. When I try that, the rescue disc scans for Fedora installations
> and does not find any, therefore it does not mount any. Examining the
> logical volume group with the xwindows LVM tool shows that it is present on
> /dev/hdb2, which is correct, and it has a different name now, VolGroup01.
> The physical view shows 189Gb of space which is correct. The logical view
> shows all of that space as free. Now I know that all my files are still
> there, the darned thing did not run long enough to erase all of the files,
> but something happened to the logical drive and now all the space is marked
> as "Free" with no files present. Bummer.

Not sure how webmin does do, but it seems to have destroyed the old "volume
group" and "logical volume" and then created the new one, this shouldn't have
included any formate.

One way that may fix this, si to crate a new "logical volume" to your new
"volume group", do this manually, add all the free space at once and then run
fsck on the new device, if you are lucky, you will after that be able to
access all your old files (don't forget to mount it after fsck).


The reason why you got a busy message when you allowed the rescue disk to
mount the system was that /boot was mounted into the system root,

umount /mnt/sysimage/boot
umount /mnt/sysimage

That had made you to unmount the logical volume.


Okey, time for me to sleep again...


//Aho
From: noi on
On Mon, 02 Oct 2006 02:43:29 +0200, J.O. Aho wrote this:

> Ohmster wrote:
>> noi <noi(a)siam.com> wrote in news:EeXTg.6607$TV3.5327
>> @newssvr21.news.prodigy.com:
>>
>>> Second do not panic.
>>> With FC3 as the primary system, boot from the FC5 DVD
>>>
>>> boot: linux rescue
>>> # you get the prompt
>>>
>>> $ chroot /mnt/sysimage
>>>
>>> $ ls ~
>>>
>>> # you should still see your data
>>>
>>> If you can see your data then you can disconnect the FC3 and reinstall
>>> FC5 without the LVM.
>>
>> No can do. When I try that, the rescue disc scans for Fedora
>> installations and does not find any, therefore it does not mount any.
>> Examining the logical volume group with the xwindows LVM tool shows that
>> it is present on /dev/hdb2, which is correct, and it has a different
>> name now, VolGroup01. The physical view shows 189Gb of space which is
>> correct. The logical view shows all of that space as free. Now I know
>> that all my files are still there, the darned thing did not run long
>> enough to erase all of the files, but something happened to the logical
>> drive and now all the space is marked as "Free" with no files present.
>> Bummer.
>
> Not sure how webmin does do, but it seems to have destroyed the old
> "volume group" and "logical volume" and then created the new one, this
> shouldn't have included any formate.
>
> One way that may fix this, si to crate a new "logical volume" to your new
> "volume group", do this manually, add all the free space at once and then
> run fsck on the new device, if you are lucky, you will after that be able
> to access all your old files (don't forget to mount it after fsck).
>
>
> The reason why you got a busy message when you allowed the rescue disk to
> mount the system was that /boot was mounted into the system root,
>
> umount /mnt/sysimage/boot
> umount /mnt/sysimage
>
> That had made you to unmount the logical volume.
>
>
> Okey, time for me to sleep again...
>
>
> //Aho

Yes, I agree with Aho. IMO Webmin thought you wanted to create a new
physical volume using hdb and wrote to the partition messing up your
existing partitions. That's the reason I make a partition for
/home separate from /boot, /swap or / partitions.

You can try booting the FC5 as primary then lvscan the FC3 on /dev/hdb
or mounting the FC3 /dev/hdb1 as /mnt/fc3sys then copy your files

You could also try a Knoppix LiveCD which should
automount all of your partitions /dev/hda1, /dev/hdb1 as read-only
/mnt/hda1, /mnt/hdb1. Then you could cp -a /mnt/hdb1/home/dirs
/mnt/hdb1/home copy from the FC3 mounted drive to the FC5 mounted drive.
If you can read the FC3 files under Knoppix. I suggest Knoppix because
it auto mounts all available hds to read-only mount points.


If you can read the files under Knoppix you change the properties of the
/mnt/hda1 from read-only to allow you to copy the files from /mnt/hdb1

Make sure you do not change properties unless you can see all the files
and permissions on FC3 and FC5. After the copies complete make sure the
permissions of the files copied to the FC5 /mnt/hda1/home/dirs have the
correct permissions, userid and groupids.

Very important to verify copied files have the right permissions, userid,
and groupid. Also very important not to change files, partitions or
anything on the FC3 drive until you can backup your data or the FC5
drive is working with all your data.

From: Ohmster on
noi <noi(a)siam.com> wrote in
news:ZD1Ug.18706$Ij.5243(a)newssvr14.news.prodigy.com:

> Yes, I agree with Aho. IMO Webmin thought you wanted to create a new
> physical volume using hdb and wrote to the partition messing up your
> existing partitions. That's the reason I make a partition for
> /home separate from /boot, /swap or / partitions.
>
> You can try booting the FC5 as primary then lvscan the FC3 on
> /dev/hdb or mounting the FC3 /dev/hdb1 as /mnt/fc3sys then copy your
> files

This is FC5, the old FC3 (What is left of it.) is on /dev/hdb. Partition
1 on hdb is the boot partition, partition 2 on hdb is the LVM where my
stuff is at:

[root(a)ohmster ~]# lvscan
ACTIVE '/dev/VolGroup00/LogVol00' [184.22 GB] inherit
ACTIVE '/dev/VolGroup00/LogVol01' [1.94 GB] inherit
[root(a)ohmster ~]# fdisk -l /dev/hda

Disk /dev/hda: 200.0 GB, 200049647616 bytes
255 heads, 63 sectors/track, 24321 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hda1 * 1 13 104391 83 Linux
/dev/hda2 14 24321 195254010 8e Linux LVM
[root(a)ohmster ~]# fdisk -l /dev/hdb

Disk /dev/hdb: 203.9 GB, 203928109056 bytes
255 heads, 63 sectors/track, 24792 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hdb1 * 1 13 104391 83 Linux
/dev/hdb2 14 24792 199037317+ 8e Linux LVM
[root(a)ohmster ~]#


> You could also try a Knoppix LiveCD which should
> automount all of your partitions /dev/hda1, /dev/hdb1 as read-only
> /mnt/hda1, /mnt/hdb1. Then you could cp -a /mnt/hdb1/home/dirs
> /mnt/hdb1/home copy from the FC3 mounted drive to the FC5 mounted
> drive. If you can read the FC3 files under Knoppix. I suggest
> Knoppix because it auto mounts all available hds to read-only mount
> points.

You know, knoppix might not be a bad idea. I will try it later, have to
go to work now.

> If you can read the files under Knoppix you change the properties of
> the /mnt/hda1 from read-only to allow you to copy the files from
> /mnt/hdb1
>
> Make sure you do not change properties unless you can see all the
> files and permissions on FC3 and FC5. After the copies complete make
> sure the permissions of the files copied to the FC5
> /mnt/hda1/home/dirs have the correct permissions, userid and groupids.

I hear ya.

> Very important to verify copied files have the right permissions,
> userid, and groupid. Also very important not to change files,
> partitions or anything on the FC3 drive until you can backup your data
> or the FC5 drive is working with all your data.
>

Will try that when I get home later. Thanks for the advice.

--
~Ohmster
theohmster at comcast dot net
Put "messageforohmster" in message body
to pass my spam filter.
From: noi on
On Mon, 02 Oct 2006 12:39:54 +0000, Ohmster wrote this:

> noi <noi(a)siam.com> wrote in
> news:ZD1Ug.18706$Ij.5243(a)newssvr14.news.prodigy.com:
>
snip
>
> This is FC5, the old FC3 (What is left of it.) is on /dev/hdb. Partition 1
> on hdb is the boot partition, partition 2 on hdb is the LVM where my stuff
> is at:
>
> [root(a)ohmster ~]# lvscan
> ACTIVE '/dev/VolGroup00/LogVol00' [184.22 GB] inherit

ACTIVE '/dev/VolGroup00/LogVol01' [1.94 GB] inherit

> [root(a)ohmster ~]# fdisk -l /dev/hda
>
> Disk /dev/hda: 200.0 GB, 200049647616 bytes 255 heads, 63 sectors/track,
> 24321 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes
>
Device Boot Start End Blocks Id System

/dev/hda1 * 1 13 104391 83 Linux

/dev/hda2 14 24321 195254010 8e Linux LVM

[root(a)ohmster ~]# fdisk -l /dev/hdb
>
> Disk /dev/hdb: 203.9 GB, 203928109056 bytes 255 heads, 63 sectors/track,
> 24792 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes
>
Device Boot Start End Blocks Id System

/dev/hdb1 * 1 13 104391 83 Linux

/dev/hdb2 14 24792 199037317+ 8e Linux LVM

[root(a)ohmster ~]#
>
>
snip
> You know, knoppix might not be a bad idea. I will try it later, have to go
> to work now.
>
Yeah but make sure you don't change file permissions, uids, gids on either
FC3 or FC5. Major headaches if that happens..
>
snip
>
> I hear ya.
>
snip
> Will try that when I get home later. Thanks for the advice.

You can see the hdb partitions using fdisk on FC5
Are you sure you can't mount the hdb2 now on FC5?
I mean case closed if you can mount FC3 on FC5

root@ $ mount -t ext3 -o ro,defaults /dev/hdb2 /mnt/fc3sys


I didn't mention it first because using
$ vgextend volgroup /dev/hdb2
should have added FC3 as an extension to your FC5 volgroup
but you should be able to turn off lvm vgchange -an.

$ man vgchange

root@ $ vgchange -an # turns off lvm

root@ $ shutdown -r now

root@ $ mount -t ext3 -o ro,defaults /dev/hdb2 /mnt/sysfc3

root@ $ cp -a /mnt/sysfc3/whatever /home/whatever

root@ $ vgchange -an # turns lvm back on

root@ $ shutdown -r now




From: J.O. Aho on
noi wrote:

> Are you sure you can't mount the hdb2 now on FC5?
> I mean case closed if you can mount FC3 on FC5
>
> root@ $ mount -t ext3 -o ro,defaults /dev/hdb2 /mnt/fc3sys

Thats not possible, you need to use the LVM device name and not the device
name of the slice where the LVM is located.


//Aho