Proxmox 2.3 - 3.0 failed Upgrade (LVM/Grub Errors)

[gavinlew]

** Reposting as I realised posted to an EOL section of the forum.

Hi All,

I would appreciate some help with my failed upgrade of Proxmox 2.3 -> 3.0 on a HP Microserver N40L (installed using 2.3 ISO directly to first Hard Disc).

Upon reboot I am being dumped into Grub and can get the system partially booting by passing boot/init commands manually via the CLI.

However, boot is failing and I am being dumped to a recovery console

Found volume group 'pve' using metadata type lvm2
3 logical volume(s) in volume group 'pve' now active
Activating lvm and md swap...done
Checking file systems...fsck from util-linux 2.20.1
fsck.ext3: No such file or directory while trying to open /dev/pve/data
Possibly non-existant device?
...
...
fsck died with exit status 8
failed (code 8).

pvscan shows PV /dev/sda2 VG pve lvm2 232.38 GiB / 16.00 GiB free]
Total: 1

lv scan shows
ACTIVE /dev/pve/swap 4.0GiB inherit
ACTIVE /dev/pve/root 58.0GiB inherit
ACTIVE /dev/pve/data 154.39GiB inherit

How can I get fsck to agree that the disc is fine, so it can mount the volumes correctly and finally grub2 to let the system boot

Any help appreciated,

Gavin

 

[gavinlew]

Noone?

Should I just mount /dev/pve/data which contains backups and get them off the machine, reformat with a new install of proxmox3 and re-add a backup?

 

[kobuki]

At first glance this looks like a case of a misconfigured/incompletely installed bootloader (grub2) and/or not properly updated initrd. Boot from a Wheezy live rescue CD and fix it -- there's really plenty of information on the net regarding this matter on Debian.

 

[gavinlew]

Spent another 4 hours on the machine trying to recover it, now im getting mismatched libraries, so using grml to extract the backups of the VM's reformatting and reinstalling which will be a lot quicker

 

[udo]

Hi,
reinstall is perhaps the faster way, but what show follwoing commands:

pvs
vgs
lvs

Udo

 

[gavinlew]

** Reposting as I realised posted to an EOL section of the forum.

Hi All,

I would appreciate some help with my failed upgrade of Proxmox 2.3 -> 3.0 on a HP Microserver N40L (installed using 2.3 ISO directly to first Hard Disc).

Upon reboot I am being dumped into Grub and can get the system partially booting by passing boot/init commands manually via the CLI.

However, boot is failing and I am being dumped to a recovery console

Found volume group 'pve' using metadata type lvm2
3 logical volume(s) in volume group 'pve' now active
Activating lvm and md swap...done
Checking file systems...fsck from util-linux 2.20.1
fsck.ext3: No such file or directory while trying to open /dev/pve/data
Possibly non-existant device?
...
...
fsck died with exit status 8
failed (code 8).

pvscan shows PV /dev/sda2 VG pve lvm2 232.38 GiB / 16.00 GiB free]
Total: 1

lv scan shows
ACTIVE /dev/pve/swap 4.0GiB inherit
ACTIVE /dev/pve/root 58.0GiB inherit
ACTIVE /dev/pve/data 154.39GiB inherit

How can I get fsck to agree that the disc is fine, so it can mount the volumes correctly and finally grub2 to let the system boot

Any help appreciated,

Gavin

Hi,
reinstall is perhaps the faster way, but what show follwoing commands:

pvs
vgs
lvs

Udo

Hi Udo,

pvs :
/dev/sda2 pve lvm2 a-- 232.38g 16.00g

vgs :
pve 1 3 0 wz--n- 232.38g 16.00g

lvs :
data pve -wi-a--- 154.39g
root pve -wi-ao-- 58.00g
swap pve -wi-a--- 4.00g

As far as I can see the partitions etc are all fine, they just dont mount properly and more

 

[udo]

Hi Udo,

pvs :
/dev/sda2 pve lvm2 a-- 232.38g 16.00g

vgs :
pve 1 3 0 wz--n- 232.38g 16.00g

lvs :
data pve -wi-a--- 154.39g
root pve -wi-ao-- 58.00g
swap pve -wi-a--- 4.00g

As far as I can see the partitions etc are all fine, they just dont mount properly and more

Hi,
and if you do an "vgchange -a y pve" (but is allready active now) you see any devices in /dev/mapper?

ls -lsa /dev/mapper
dmsetup ls --tree

Udo

 

[gavinlew]

Hi,
and if you do an "vgchange -a y pve" (but is allready active now) you see any devices in /dev/mapper?

ls -lsa /dev/mapper
dmsetup ls --tree

Udo

Hi again Udo,

If i type
# vgchange -a y pve
3 logical volume(s) in volume group pve now active
# ls -lds /dev/mapper
Only one entry
0 crw------T 1 root root 10, 61 Jun 25 22:05 control
# dmsetup ls --tree
pve-swap 253:0
8:2
pve-root 253:1
8:2
pve-data 253:2
8:2

Does this help?

I have booted the system with grml64 and can mount the existing pve-data LVM , I was just going to copy the vm's from the dump folder to another HDD then reinstall , or do you think the existing install can be saved?

TIA,
Gavin

 

[udo]

Hi again Udo,

If i type
# vgchange -a y pve
3 logical volume(s) in volume group pve now active
# ls -lds /dev/mapper
Only one entry
0 crw------T 1 root root 10, 61 Jun 25 22:05 control
# dmsetup ls --tree
pve-swap 253:0
8:2
pve-root 253:1
8:2
pve-data 253:2
8:2

Does this help?

I have booted the system with grml64 and can mount the existing pve-data LVM , I was just going to copy the vm's from the dump folder to another HDD then reinstall , or do you think the existing install can be saved?

TIA,
Gavin

Hi,
is on your installation udev running?

ps aux | grep udev

Do you see the disks here:

ls -lsa /dev/disk/by-id/*pve*

On the other way is an new install an fast (and clean) option. ;-)


Udo

 

[kobuki]

ls -d lists the given parameter as itself, when the parameter is a directory. It'd be better like ls -la /dev/mapper/. grml is supposedly a good rescue system, but I'm not sure the latest one already uses wheezy. It'd be best to use a genuine Debian 64-bit live rescue system to make sure your subsystems, modules and binaries match that of the Proxmox 3.0. The vgchange -ay command is supposed to activate the VG and all its LVs, do you have a /dev/pve/ directory after running it? If not it would lead to a very lengthy rescue session, best would be probably a reinstall/restore combo. Also, adding -vvv as parameter to the lvm commands might give some additional clues.