Hi all,
Hopefully this is the right forum for this,
I have been a very happy Qubes 4 (which uses LVM thin pools) user for some time now and have had no issues until the other day. I stupidly allocated a VM too much space and totally filled the volume group, which crippled Qubes. When I would login I couldn't even open the VM manager, let alone the backup.
I have tried fixing the problem by myself using a Linux Mint live-USB (with much googling), and have added a new pv to the vg, but for some reason was not able to expand the vg, move the vg, or even use any command such as lvconvert --repair, vgchange. It turns out that lvm thinks one of the pv is missing (my luks partition), and this stops any attempt of repair, even when using --partial, etc.
I then realised when looking in /etc/lvm/archive that the vg file is listing my first pv as an unknown device with a missing flag. I am figuring this is a result of me having to mount the partition using cryptopen to /dev/mapper/qubes. I subsequently corrected the path and removed the missing flag (using this guide -
https://mbroz.fedorapeople.org/talks...vmrecovery.pdf). I then go to restore the corrected .vg file (sudo vgcfgrestore -f /etc/lvm/archive/thevgfile.vg my_vg.. and get told that it cannot restore the vg with thin logical volumes as it is not supported yet and thus the restore failed.
I have tried googling my way out of this once again, but I'm afraid I am stuck. I know the pv's are all fine, and the drives are working well (apart from two of the pv being totally full).
Am I out of luck with this, or is there a way of correcting the missing vg with thin logical volumes?