Proxmox not booting after Upgrade

[Moritz Weichert]

Hey guys,

after upgrading the kernel from 4.10.15 to 4.10.17 i ended up with an non booting system.
During boot i get the message, that the system cannot connect to lvmetad and that the superblock for /dev/mapper/pve-root cannot be found. My system is running on an usb device by the way.
I didn't found very much information on what i can do. I have an 8TB data Disk (zfs) and an 512GB vm ssd (zfs). So would the fastest thing be to reinstall the usb and is it then possible to reintegrate all the lxc container and the data? No encryption so far.

Would be appreciate for fast help.

Regards
Moritz

 

[RedneckBob]

"So would the fastest thing be to reinstall the usb"

I read somewhere you need to set the boot delay to a minute so the LVM can start up if you insist from running Proxmox off a USB stick, but I'm not sure if that is the case or not. Generally it is advised against running Proxmox from a USB stick so I have no experience with your exact issue.

Is /etc on this USB stick? Is this machine part of a cluster or a node? Do you have backups?

A more permanent and durable fix may be to re-install Proxmox on a real harddrive (a small SSD would be fine) and restore/rebuild your VMs. If you go that route, unplug your 8T and 512GB disk drives, install Proxmox on a harddrive, and once you're done installing, reboot, and then recreate all your VMs locally but don't make them bigger than a few megabytes and set them to raw. Once that is done make sure all the VMs are powered off and "start at boot = NO" for each VM. Power down the server.

Plug in your 8T, and 512G drives, boot up, and go to the command line to run "zpool import". The zpool on your 8T and 512GB should import. Immediately snapshot the filesystems that contain your VMs and important data. You could "zfs send" a copy to another machine.

Now you have to go to /etc/pve/nodes/$HOSTNAME/qemu-server and edit each configuration file for your VMs to replace the line that looks like this:

virtio0: zfs-local:vm-100-disk-1,discard=on,size=32G

With one that matches the location of your VM on the 500GB drive. Maybe something like this:

virtio0: zfs-pool:vm-100-disk-1,discard=on,size=32G

Try to start each VM, debug any issues that appear, and then once everything is running throw the USB stick in the bin.

 

[manu]

@Moritz Weichert
do you get the same situation as in http://blog.wittchen.biz.pl/ubuntu-system-boot-problem/
then do vgchange -ay , then exit

if this helps, then add a rootdelay paramater to your command line as explained in
http://blog.wittchen.biz.pl/ubuntu-system-boot-problem/ ( attempt 1)
don't forget to call update-grub afterwards